squeezing light in nanoparticle-film plasmonic metasurface

Thèse de doctorat

de l’UTT

Rana Nicolas

Squeezing Light in Nanoparticle-film Plasmonic Metasurface:

from Nanometric to Atomically Thin Spacer

Spécialité : Optique et Nanotechnologies

2015TROY0028 Année 2015

Thèse en cotutelle avec l’Université Libanaise - Beyrouth - Liban

THESE

pour l’obtention du grade de

DOCTEUR de l’UNIVERSITE DE TECHNOLOGIE DE TROYES

Spécialité : MATERIAUX, MECANIQUE, OPTIQUE ET NANOTECHNOLOGIE

présentée et soutenue par

Rana NICOLAS

le 20 octobre 2015

Squeezing Light in Nanoparticle-film Plasmonic Metasurface : from Nanometric to Atomically Thin Spacer

JURY

M. M. CANVA DIRECTEUR DE RECHERCHE CNRS Président Mme M. ABBOUD MAITRE DE CONFERENCES Rapporteur M. Z. HERRO PROFESSEUR Directeur de thèse M. T. MAURER MAITRE DE CONFERENCES Directeur de thèse M. J. R. KRENN PROFESSOR Rapporteur M. M. TABBAL FULL PROFESSOR Examinateur

Personnalités invitées M. P.-M. ADAM PROFESSEUR DES UNIVERSITES M. G. LÉVÊQUE MAITRE DE CONFERENCES

Acknowledgments

First, I would like to express my sincere gratitude to my supervisors. Dr Thomas

Maurer for the continuous support during my Ph.D study, for his patience, motiva-

tion, enthusiasm, knowledge, and most importantly for his trust allowing me a wide

margin of independence to grow as a young researcher. One simply could not wish

for a better or friendlier supervisor. And Dr Ziad Herro for all his help, support,

and knowledge which made this PhD thesis a pleasant journey.

I would also like to thank my thesis committee, Dr Joachim Krenn, Dr Marie

Abboud, Dr Bert Hecht, Dr Malek Tabbal, Dr Michael Canva for accepting the

invitation to be part of the committee and for their insightful comments and en-

couragement which highly enriched my thesis.

Another big thank you goes to Dr Gaëtan Lévêque, for all the help he provided

with numerical simulations, the impact of these simulations and the fruitfull long

discussions we had was crucial to complete this work. I am also very thankful to

Dr Pierre Michel Adam for all the guidance and support he offered since the very

beginning of my PhD, and for all the scientific discussions we had which highly

enhanced my knowledge on plasmonics, his expertise was essential in understanding

the different experimental results. Thanks to Dr Michel Kazan for all his support,

especially at the early stages of my PhD.

Thanks to the hard workers in the lab, Sergei Kostcheev, Régis Deturche, and

Jeremie Beal, for all the training sessions and technical assistance which helped me

develop the necessary skills to finish this work. And thanks to all the members and

colleagues of the LNIO laboratory, because of you i was lucky enough to work in a

friendly and productive environment. A special thank you goes to Nancy Rahbany,

our friendship have definitely made my stay in France, and in Troyes much more

pleasant, and helped me throughout the difficult times of my PhD. Thanks to all

my Lebanese friends, one could never feel away from home when surrounded by you.

A very special thanks goes to my beloved parents, their unlimited love, support

and encouragement made me the person i am today, i owe this success to you. And

to my brother, sister, and little John thank you for always being my source of joy.

I would like to dedicate this thesis to my brother Dany’s soul, he was always

my inspiration and the reason i love science.

1

Contents

Acknowledgments 1

General Introduction 13

1 Glamour of plasmonics 171.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1.2 Plasmonics going down to the nanoscale . . . . . . . . . . . . . . . . 17

1.2.1 Bulk Plasmons . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.2.2 Plasmon resonance in metallic film . . . . . . . . . . . . . . . 20

1.2.3 Metallic nanoparticles and their plasmonic properties . . . . . 23

1.3 In depth view of different plasmonic modes . . . . . . . . . . . . . . . 25

1.3.1 Localized modes in nanospheres . . . . . . . . . . . . . . . . . 26

1.3.2 Localized modes in nanorods . . . . . . . . . . . . . . . . . . . 26

1.3.3 Coupling between neighboring metallic nanoparticles . . . . . 27

1.3.4 More complex geometries and the hybridization model . . . . 29

1.3.5 Symmetric and assymetric delocalized plasmon modes in IMI

and MIM structures . . . . . . . . . . . . . . . . . . . . . . . 30

1.3.6 Gap modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

1.3.7 The dark side of plasmonics . . . . . . . . . . . . . . . . . . . 33

1.3.8 Fano modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1.4 Optical properties of film-nanoparticles coupled systems . . . . . . . . 39

1.4.1 Breakthroughs in the physical understanding of coupled systems 40

1.4.2 Coupling between ordered arrays of metallic nanoparticles and

metallic film . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

1.4.3 Potential applications in NP-film systems . . . . . . . . . . . . 42

1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

2 Methods and techniques 472.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2.2 Fabrication techniques . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2.2.1 Thin film deposition . . . . . . . . . . . . . . . . . . . . . . . 47

2.2.2 Fabrication of Au NPs Via Electron Beam Lithography (EBL) 48

2.3 Structural characterization . . . . . . . . . . . . . . . . . . . . . . . . 51

2.3.1 Scanning Electron Microscopy (SEM) . . . . . . . . . . . . . . 51

2.3.2 Ellipsometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

2.3.3 X-Ray photo electron microscopy (XPS) . . . . . . . . . . . . 55

i

Contents Contents

2.4 Optical characterization . . . . . . . . . . . . . . . . . . . . . . . . . 56

2.4.1 Surface Plasmon Resonance (SPR) spectroscopy . . . . . . . . 56

2.4.2 UV-VIS-NIR spectroscopy . . . . . . . . . . . . . . . . . . . . 57

2.4.3 Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 60

2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

3 Plasmonic mode interferences and Fano resonances in MlM nanostruc-tured interface 633.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

3.2 Sample fabrication and structural characterization . . . . . . . . . . . 64

3.2.1 Sample Fabrication . . . . . . . . . . . . . . . . . . . . . . . . 64

3.2.2 Structural Characterization . . . . . . . . . . . . . . . . . . . 64

3.3 Influence of the metallic film on the plasmonic properties of metal NPs 66

3.4 Ultra thin 3 nm SiO2 spacer layer . . . . . . . . . . . . . . . . . . . . 71

3.5 Optical properties of MIM structure with 6 nm SiO2 spacer layer . . 72

3.5.1 Experimental results . . . . . . . . . . . . . . . . . . . . . . . 73

3.5.2 Numerical simulations . . . . . . . . . . . . . . . . . . . . . . 76

3.5.3 Angle resolved measurements . . . . . . . . . . . . . . . . . . 80

3.5.4 Understanding the Fano resonance . . . . . . . . . . . . . . . 83

3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

4 Influence of the spacer thickness on the plasmonic properties 914.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

4.2 Sample preparation and structural characterization . . . . . . . . . . 91

4.3 Evolution of Plasmonic modes versus the increase in the spacer thickness 92

4.3.1 Small NPs (diameter ≤ 110 nm) . . . . . . . . . . . . . . . . . 93

4.3.2 Large NPs (diameters≥140 nm) . . . . . . . . . . . . . . . . . 96

4.4 Effect of the Numerical Aperture . . . . . . . . . . . . . . . . . . . . 99

4.5 Narrow band super absorber . . . . . . . . . . . . . . . . . . . . . . . 102

4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

5 Graphene a sub-nanometer two dimensional spacer layer 1055.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

5.2 Understanding the Raman spectra of graphene . . . . . . . . . . . . . 106

5.3 Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

5.4 Structural characterization . . . . . . . . . . . . . . . . . . . . . . . . 108

5.5 Localized surface plasmon resonance of the graphene based coupled

NP/film interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

5.6 Interaction of graphene with metals and the tunability of the optical

properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

5.6.1 p-doping of graphene and LSPR blue shift . . . . . . . . . . . 117

5.6.2 n-doping of graphene and LSPR red shift . . . . . . . . . . . . 120

5.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

ii

Contents

6 Coupled NP/film systems for SERS and RI sensing 1256.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

6.2 Refractive index sensing . . . . . . . . . . . . . . . . . . . . . . . . . 125

6.2.1 Physical background . . . . . . . . . . . . . . . . . . . . . . . 125

6.2.2 Au NPs on graphene coated Au film/glass substrate . . . . . . 128

6.2.3 Au NPs on MIM interface . . . . . . . . . . . . . . . . . . . . 129

6.3 Surface Enhanced Raman Spectroscopy (SERS) . . . . . . . . . . . . 130

6.3.1 The electromagnetic theory of SERS . . . . . . . . . . . . . . 130

6.3.2 Surface Enhanced Raman Spectroscopy of graphene . . . . . . 132

6.3.3 Experimental results . . . . . . . . . . . . . . . . . . . . . . . 133

6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

Conclusion 139

7 French Summary 141

Bibliography 163

iii

List of Figures

1.1 Dispersion relation of the free electron gas and the dispersion of

light[1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

1.2 Surface plasmon polariton at gold/air interface . . . . . . . . . . . . . 20

1.3 Dispersion relation of surface plasmons compared to light in vacuum

and in the dielectric medium[1] . . . . . . . . . . . . . . . . . . . . . 21

1.4 excitation of SPP through prism coupling, the kretchman configuration 21

1.5 Prism coupling an SPP dispersion[2] . . . . . . . . . . . . . . . . . . 22

1.6 Grating coupling for light with a wavevector k incident on a metal

grating surface with a periodicity a . . . . . . . . . . . . . . . . . . . 22

1.7 localized surface plasmon on a metal nanoparticle in the presence of

an electromagnetic field . . . . . . . . . . . . . . . . . . . . . . . . . . 23

1.8 Charge distributions of dipole, quadrupole, hexapole surface plasmon

are depicted. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

1.9 Schematic showing the longitudinal and the transverse localized plas-

mon mode of a nanorod . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.10 Schematic of near-field coupling between metallic nanoparticles for

different polarization schemes . . . . . . . . . . . . . . . . . . . . . . 29

1.11 Schematic of plasmon hybridization in metallic nanoshells, where ws

is the frequency of the sphere, and wc the frequency of the cavity. . . 30

1.12 Symmetric and antisymmetric coupled SPP modes in a) IMI and b)

MIM structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

1.13 Longitudinal mode for a dimer with small seperation showing a hot

spot in the gap between the NPs. . . . . . . . . . . . . . . . . . . . . 32

1.14 Schematic of the generation of a hot spot in the gap between a metallic

nanoparticle and a metallic film. . . . . . . . . . . . . . . . . . . . . . 33

1.15 Full modal spectrum of a silver nanodisk with 200 nm diameter and

30 nm thickness on a 30 nm thick silicon nitride membrane. (a) Ex-

perimental (blue) and simulated (red) EEL spectra from the particle

edge (top), center (middle), and the region around half radial distance

(bottom), as indicated in red on the disk sketches.(b) Simulated sur-

face charge distributions (right) generated by assuming the electron

beam positions marked by the cross. Figure taken from ref [3] . . . . 34

1.16 Illustration of the Fano profile. Figure taken from ref[4]. . . . . . . . 35

1.17 Normalized Fano profiles with the prefactor 1/(1 + q2)for different

values of q. Figure taken from[4]. . . . . . . . . . . . . . . . . . . . . 36

1

List of Figures List of Figures

1.18 Calculated extinction spectra of non-concentric ring/disk cavity of

partial a) and complete b) filling of the cavity (yellow areas in the in-

set) with a dielectric medium of permittivity 1 (black solid), 1.5 (blue

dashed) and 3 (red dotted). The structure is placed on a glass sub-

strate modelled using a permittivity of 2.04, c) extinction spectra of

thenon-concentric ring/disk cavity as a function of angle of incidence

θ defined in the inset. The charge amplitude plots in the panels to

the right show the symmetry of the higher multipolar modes at their

resonance wavelengths. Figure taken from ref [5]. . . . . . . . . . . . 37

1.19 a) Calculated dipole amplitudes of the bonding and antibonding col-

lective dipolar plasmon modes in a gold nanoshell heptamer [6], b)

Transmission spectra showing the effects of coupling in lithographi-

cally fabricated gold nanodisk heptamers. Figure taken from ref [7]. . 38

1.20 a) Schematic of interface reflections and a summary of the Fano reso-

nance interference conditions, b) Specular reflection spectra for differ-

ent angle of incidence. The assymetric line-shapes is for angles close

to the critical angle of the glass-air interface, θc = 41.1° . Dotted

lines are experimental data while the solid lines are fits to the Fano

formula. Figure taken from ref [8]. . . . . . . . . . . . . . . . . . . . . 39

1.21 a) Geometry of the sample studied by Hao et al b) SEM images of

the NPs c) Measured and simulated absorbtion spectrafor Wx=170

nm, Wy=230 nm, t =40 nm, d= 10nm, h=50nm, and a = 30 nm at

20° incidence angle. Figure taken from ref[9]. . . . . . . . . . . . . . . 43

1.22 (a) Geometry of the sample studied, Au film with Ag nanocubes

separated by a 10nm polymer spacer layer (n=1.54) (b) SEM image

of the fabricated optical metamaterial absorber. (c) Experimental

reflectance for normal incidence, normalized to the gold film, for sur-

face coverages of 7.3% (thin solid line) and 17.1% (thick solid line),

compared with numerical simulations of uniform cubes (4.2% surface

coverage, dotted line) and a model including size dispersion. Figure

taken from ref [10]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

1.23 (a) Geometry of the sample, the unit cell is quadratic of size ▲ having

four Au NPs of different sizes. (b) Schematic representation of the

layered sample structure. (c) SEM image of the fabricated GPRs

with particle diameters d1 = 60nm, d2 = 80nm, d3 = 100nm and d4

= 120nm and unit-cell size ▲ = 340nm. (d) Reflection spectra for

unpolarized (solid red curve), x-polarized (dashed blue curve) and y-

polarized (black dash-dotted curve) light. Figure taken from ref [11].

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

2

List of Figures

2.1 Schematic of the elctron beam lithography. 1) Spin coating the glass

substrate with a thin layer electron sensitive resist. 2) Spin coating

the sample with a conductive polymer to avoid the charging effect. 3)

Imprinting the designed pattern with the electron beam. 4) Rinsing

with water to remove the conductive polymer. 5) Development with

MIBK/IPA. 6) Evaporation of a 3 nm adhesive Chromium layer. 7)

Evaporation of metal. 8) Lift off using acetone as a solvant to remove

the remaining resist and the metal on top of it, resulting in metal

nanostructures on the desired substrate. . . . . . . . . . . . . . . . . 49

2.2 Illustration of a scanning electron microscope (SEM) showing the

different component. Image courtesy of Iowa state University. . . . . 53

2.3 Schematic of an ellipsometer showing the light source, the polarizer,

the compensator, the sample holder, the analyzer and the detector

which measures the intensity transmitted from the analyzer. . . . . . 54

2.4 Schematic of the steps required to perform ellipsometry measurements. 55

2.5 Schematic showing the main components of an X-Ray phto electron

microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

2.6 Schematic of SPR spectrometer in wavelength scan mode. . . . . . . 57

2.7 Scematic of the home made extinction spectrometer [12]. . . . . . . . 59

2.8 Scematic of the TE2000-U microscope, showing the different com-

ponents of the microscope. The T-DH dia illuminator 100W, LHS-

H100P-1 12V100W halogen lamp, TE2-PS100W power supply, a rect-

angular stage used as a sample holder, a binocular eyepiece tube, a

system condenser, a 0.45 objective, cords ... . . . . . . . . . . . . . . 60

2.9 Quantum Energy Transitions for Rayleigh and Raman Scattering. . . 61

3.1 Schematic of the designed interfaces with Au NPs on top of a) glass

substrate, b) 50 nm Au film on a glass substrate and c) SiO2 layer/Au

film on a glass substrate. . . . . . . . . . . . . . . . . . . . . . . . . 65

3.2 Dark field microscopy image showing twelve Au arrays, the x-axis

displays the diameters of the NPs in the arrays: 200, 170, 140, 110

nm, and the y-axis displays the three different periodicities 300, 450

and 600 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

3.3 SEM images of a) full array, b-d) higher magnification showing the

different periodicities e-i) parts of the arrays with 300 nm periodicity

and different NP diameters, j-n) single NP with different diameter . . 67

3.4 Extinction spectra of NPs with five different diameters: 80, 110, 140,

170, and 200 nm deposited on a) glass substrate b) 50 nm Au coated

glass substrate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.5 Resonant wavelength versus the change in the diameter for two dif-

ferent periodicities: 300 nm (red) and 450 nm (green) for the a) glass

substrate, and b) Au film. . . . . . . . . . . . . . . . . . . . . . . . . 69

3


3.6 Computed distribution of the electric field inside a vertical section of

a 200 nm Au NP on a 50 nm Au substrate at a wavelength λ= 520

nm. The green vectors show the real part of the electric field. . . . . 70

3.7 Extinction spectra under normal incidence for Au NPs with 300 nm

periodicity and five different diameters on a 3nm SiO2/ 50 nm Au film. 72

3.8 Extinction spectra in transmission and under normal incidence for

different diameters of Au NPs (80, 110, 140, 170 and 200 nm) with

center-to-center distances of a) 300 nm, b) 450 nm and c) 600 nm. . 74

3.9 Resonance wavelength versus the diameter of NPs for the three differ-

ent periods (300, 450 and 600 nm): a) for the localized mode at low

wavelengths around 520 nm, b) for the delocalized mode at higher

wavelengths between 560 and 620 nm and c) for the Fano resonance. . 75

3.10 (a) Absorption spectra of a single cylinder gold particle of varying

diameter on top of the multilayer substrate. The excitation is a TM-

polarized plane wave in normal incidence, from the air side. (b) Dis-

tribution of the electric-field intensity for modes indicated on spectra

(a): the distribution is plotted in the polarization plane for modes

(1) and (d), and in a plane parallel to the substrate just under the

bottom surface of the cylinder for modes (2) and (3). Maps with

the same label on the spectra (a) have similar field distributions; the

green arrows on maps (1) and (d) show the real part of the electric

field. (c) Amplitude of the field scattered at infinity in the direction

of the transmitted incident wave (0°) for particles with diameters 80

nm (black) and 200 nm (green), which slowly vanishes asymptotically. 77

3.11 (a) Evolution with the period of the excitation wavelength of the two

PSP supported by the substrate: black: air-side PSP, red: silica-side

PSP. The incident planewave arrives normally to the interface, from

the air. On the right are plotted the profile of the electric field for

different wavelengths indicated by the numbers. (b) Folded dispersion

curves of the two PSP modes associated to the substrate, for a period

of 300nm. The wavelength ❧ in vacuum varies from 500nm to 900nm,

k0=2♣/❧ and kx is the parallel component of the incident wavevector. 79

3.12 . Evolution of extinction spectra versus illumination and collection

angles for a) TE and b) TM illumination. . . . . . . . . . . . . . . . . 80

4

List of Figures

3.13 (a) Extinction spectra of a 2D square grating of 3D gold nanopar-

ticles of 50 nm thickness and 200 nm diameter on the multilayered

substrate, period 300nm and TM polarization. The solid black and

red lines are eye guides for the evolution of the SPP modes on air

(black) and glass (red) side, the dashed lines corresponds to extinc-

tion spectra for angles with 5° step between the main angles. The

vertical green lines indicate the position of the non-dispersive gap-

modes. The distribution of the electric field amplitude inside the

polarization plane has been plotted for few selected angles and wave-

lengths indicated in the spectra; the green arrows are the real part of

the electric field. (b) Comparison between the extinction spectra for

TM and TE excitation as a function of the incidence angle. . . . . . . 82

3.14 Schematic of a highly relective substrate, showing both the reflection

and transmission. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

3.15 Evolution with the wavelength of the reflection coefficient on the ex-

perimental substrate with the 6nm-silica/50nm-gold (dashed line) and

for a semi-infinite gold substrate (solid line). . . . . . . . . . . . . . 85

3.16 the extinction spectrum associated to the first, long-wavelength reso-

nance Im (α1(ω)) in blue, the extinction spectrum associated to the

second, short-wavelength resonance Im(α2(ω)) in red, and the extinc-

tion spectrum associated with the bi-resonant particle Im(α(ω)) in

black. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

3.17 a) the reflected and the transmitted scattered field b)the amplitude

of the scattered fields, both proportional to abs(❛). As in the full

Green’s tensor simulation. . . . . . . . . . . . . . . . . . . . . . . . . 87

3.18 Theoretical reflection and transmission plots scaling as Im(α) and

Re(α) respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

3.19 Experimental a) transmission and b) reflection measurements under

normal incidence for a 4.5 nm SiO2 MIM structure. . . . . . . . . . . 88

4.1 a) Experimental and b) numerical extinction spectra for NPs grating

with a diameter of 80 nm and a periodicity of 300 nm, deposited on

Au film with an SiO2 spacer layer with thickness varying from 10 up

to 100 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93




to 100 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94




to 100 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

5





to 100 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

4.5 a) Experimental and b) Numerical (with an angle average to account

for the NA) extinction spectra for NPs of different diameters and a

periodicity of 300 nm on an MIM structure with 100 nm SiO2spacer

layer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

4.6 Numerical simulations showing the evolution of the extinction spectra

with θ, for different fixed values of ϕ, for a spacer thickness of 100 nm.101

4.7 Experimental extinction spectra under TM polarization for NPs of

different diameters and a) 300 nm and b) 450 nm periodicity. . . . . . 103

5.1 Raman spectra of a) perfect graphene monolayer, and b) disordered

graphene layer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

5.2 Schematic illustration of the fabrication steps of a graphene-based

SPR interfaces decorated with an ordered array of gold NPs using

Electron Beam Lithography. . . . . . . . . . . . . . . . . . . . . . . . 109

5.3 a) Raman spectra of the transferred graphene (blue) and the original

graphene on Ni (red); b) C1s high resolution XPS spectra of graphene

before (solid red) and after (dashed blue) transfer. . . . . . . . . . . . 110

5.4 Surface plasmon resonance curves of Au surfaces before (black) and

after (blue) transfer of graphene: experimental data (dotted lines);

theoretical SPR curves (full lines) [nprism= 1.52, nT i = 2.36+i3.47

(d=5 nm); nAu = 0.197+i3.67 (d=50 nm); nAg = 0.14+i4.581 (d=38

nm); ngraphene = 3.0+i1.216; l=670 nm. . . . . . . . . . . . . . . . . . 111

5.5 Raman spectroscopy a) provided by ACS materials and b) measured

in our lab on graphene on quartz substrates, showing a good quality

monolayer graphene. . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

5.6 SEM images of graphene-based SPR decorated with Au NPs by EBL

with center-to-center distnace of 300 nm. The particles are 50 nm in

height and 80 nm (a), 110 nm (b), and 140 nm (c) in diameter. . . . 112

5.7 Schematic illustration of the different interfaces investigated: a graphene-

coated Au thin film decorated with Au NPs array; b Au NPs array

directly deposited onto thin Au film without the graphene spacer layer.113

5.8 Extinction spectrameasured in air of the Au surface (dashed lines)

and graphene-modified Au surface (full lines) decorated with Au NPs

of 50 nm (black), 80 nm (blue), 110 nm (green), and 140 nm (red)

in diameter, 50 nm in height and center-to-center distance of 300

nm. The signal was collected with a ×10 objective with a numerical

aperture of NA=0.15. The reference for calculating the extinction

is taking on the gold film outside the arrays. The optical extinction

spectrum of 80 nm Au NPs directly fabricated on Au film could not

be resolved. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

6

List of Figures

5.9 Computed extinction spectra of a single cylinder particle, diameter

80 nm (blue), 110 nm (green), and 140 nm (red), thickness 50 nm,

placed on the Au surface (solid lines), or on the graphene-modified

Au surface (dashed lines). . . . . . . . . . . . . . . . . . . . . . . . . 115

5.10 a) Computed distribution of the electric field inside a vertical section

of the 110-nm diameter particle on the Au substrate, at the reso-

nant wavelength ❧ =524 nm. Color scale electric-field time-averaged

amplitude, normalized to the incident plane wave amplitude; green

vectors: electric field real part; cyan vectors: electric field imaginary

part. b) Computed distribution of the electric field in a horizontal

section 25 nm above the Au interface, same wavelength. . . . . . . . . 116

5.11 Schematic of the cone representing the electronic structure of graphene,

the cone center is the "Dirac point"; which is equivalent to the "Fermi

level" in graphene, and the effect of n and p-doping on this level. . . . 117

5.12 Schematic of the Au NPs based interface showing Au NPs of different

dimensions fabricated by EBL on top of mnolayer graphene/quartz

substrate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

5.13 a) and b) Raman spectra showing the G peak and 2D peak respec-

tively for Au NPs/graphene quartz the black arrows are guidelines to

show the upshift of both G peak and the 2D peak when measured

on regions with Au NPs compared to thopse without Au NPs. b)

Extinction spectroscopy: red is for NPs/monolayer graphene/quartz,

and black is for NPs/quartz, a blue shift ∼13 nm is recorded for 140

and 170 nm NPs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

5.14 Schematic of the Au NPs based interface showing Au NPs of different

dimensions fabricated by EBL on top of mnolayer graphene/quartz

substrate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

5.15 a) and b) Raman spectra showing the G peak and 2D peak respec-

tively for Au NPs/graphene quartz the black arrows are guidelines

to show the upshift of the G peak and the downshift of the 2D peak

when measured on regions with Au NPs compared to those with-

out Au NPs. c) Extinction spectroscopy: red is for NPs/monolayer

graphene/quartz, and black is for NPs/quartz, a blue shift ∼13 nm

is recorded for 140, 170 and 200 nm NPs. . . . . . . . . . . . . . . . . 121

6.1 a) Extinction spectra of the graphene-modified SPR surface decorated

with Au NPs array of 140 nm in diameter for different refractive

indexes n of glycerol/water mixtures: 1.00 (black), 1.33 (gray), 1.37

(blue), 1.40 (magenta), 1.44 (red), and 1.47 (green). b) Shift of the

low wavelength LSPR peak dependingon the refractive index of the

surrounding medium. . . . . . . . . . . . . . . . . . . . . . . . . . . 128

7


6.2 a) Extinction spectra of the Au NPs with 200 nm in diameter and 300

nm in periodicity, for different refractive indexes n of glycerol/water

mixtures: 1.00 (black), 1.330 (red), 1.3697 (light blue), 1.4016 (ma-

genta), 1.4410 (green), and 1.4616 (dark blue) b) Shift of the low

wavelength LSPR peak as a function of the refractive index of the

surrounding medium; c) Numerical simulation of the extinction spec-

tra of the Au NPs with 200 nm in diameter and 300 nm in periodicity,

for different dielectric constants n2 of the NPs’ surrounding medium ;

the inset shows the distribution of the amplitude of the electric field in

the polarization plane for n=1.4 and ❧=610 nm ; d) Numerical shift of

the low wavelength localized plasmon with n2 and corresponding lin-

ear regression. The refractive index sensitivity for the low wavelength

mode is equal to 246 nm/RIU. . . . . . . . . . . . . . . . . . . . . . . 131

6.3 Raman measurements on graphene monolayer deposited on top of a

50 nm Au film, showing the enhancement for the regions with NPs

compared to those without NPs. The NPs are prepared by EBL and

are of 80 to 200 nm diameter with a periodicity equal to 300 nm. a)

represents the G-band and b) the 2D-band. . . . . . . . . . . . . . . 135

7.1 Description du procédé de lithographie électronique et des étapes as-

sociées. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

7.2 Schémas des différentes interfaces conçues et étudiées, à savoir des

NPs d ‘Au déposées sur a) des substarts de verre, b) des substrats

film d’Au/verre et c) des substrats SiO2/film d’Au/verre. . . . . . . . 147

7.3 Image obtenue en microscopie optique en champ sombre de douze

réseaux de NPs d’Au présentant des diamètres de 200nm, 170nm,

140nm, 110nm (axe x) et trois périodicités, 300nm, 450nm et 600nm

(axe y). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

7.4 Spectres d’extinction optique de NPs avec cinq diamètres différents,

80, 110, 140, 170 et 200nm déposées sur a) un substrat de verre et b)

un substrat de verre recouvert d’un film d’Au de 50nm. . . . . . . . . 149

7.5 Spectres d’extinction obtenus en transmission et sous incidence nor-

male pour différents diamètres des NPs d’Au (80, 110, 140, 170 et

200nm) avce des distances centre-à-centre de a) 300nm, b) 450nm et

c) 600nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

7.6 (a) Spectres d’absorption d’une particule d’Au cylindrique unique

pour différents diamètres déposée au-dessus d’un substrat multicouches

. L’excitation est obtenue par une onde plane incident polarisée TM

(transverse magnétique) sous incidence normale en provenance du

côté air. (b) Distribution de l’intensité du champ électrique pour

les modes visibles et indiqués sur le spectre (a). (c) Amplitude du

champ diffusé à l’infini dans la direction de l’onde incidente pour des

particules de 80nm (ligne noire) et 200nm (ligne verte). . . . . . . . . 151

8

List of Figures

7.7 Evolution des spectres d’extinction en function des angles d’illumination

et de collection pour une polarization a) TE et b)TM. . . . . . . . . . 152

7.8 Spectres d’extinction d’un réseau carré de NPs d’or de 50nm d’épaisseur,de

diamètre 200nm et de périodicité 300nm fabriquées sur un substrat

multicouches pour une polarisation incidente TM.Les lignes pleines

noires et rouges permettent de suivre l’évolution des modes SPP à

l’interface air (noir) et verre (rouge).Les lignes vertes indiquent la po-

sition spectrale des modes de gap non-dispersifs. La distribution de

l’amplitude du champ électrique à l’intérieur du plan de polarisation

a été tracée pour quelques angles et longueurs d’onde sélectionnés.

Les flèches vertes sont la partie réelle du champ électrique. (b) Com-

paraison entre les spectres d’extinction obtenus pour une polarisation

TM et une polarisation TE en fonction de l’angle d’incidence. . . . . 153

7.9 Spectres d’extinction optique a) expérimentaux et b) numériques pour

des réseaux de NPs d’Au de 80nm et de périodicité 300nm déposés

sur des films d’Au recouverts par une couche de SiO2 dont l’épaisseur

varie de 10nm à 100nm. . . . . . . . . . . . . . . . . . . . . . . . . . 154

7.10 Spectres d’extinction optique a) expérimentaux et b) numériques pour

des réseaux de NPs de diamètre de 200nm et de périodicité 300nm dé-

posés sur des films d’Au recouverts d’une couche de SiO2 d’épaisseur

variant de 10 à 100nm. . . . . . . . . . . . . . . . . . . . . . . . . . . 155

7.11 Illustration du procédé de fabrication d’un des systèmes étudiés à

savoir une interface où les NPs d’Au sont couplées à un film d’or via

du graphène comme couche séparatrice. . . . . . . . . . . . . . . . . . 156

7.12 Illustration des deux types d’interfaces étudiés: a) réseau de NPs d’or

depose directement sur un film d’or recouvert de graphene et b) réseau

similaire de NPs d’or mais deposes directement sur un film d’or, sans

presence de graphene. . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

7.13 Spectres d’extinction d’interfaces avec des NPs d’or déposées directe-

ment sur un film d’or recouvert (lignes pleines) ou non (lignes en

pointillés) de graphène. Les NPs ont un diamètre de 50nm (noir),

80nm (bleu), 110nm (vert) et 140nm (rouge) et une distance centre-

à-centre de 300nm. Le signal a été collcté avec un objectif x10 et une

ouverture numérique de 0,15. La référence a été prise sur une zone

de l’échantillon ne contenant pas de NPs. . . . . . . . . . . . . . . . . 158

7.14 Spectres Raman montrant l’évolution de a) la bande G et b) la bande

2D du graphene pour des interface graphene/quartz (noir), NPs de

140nm/graphene/quartz (rouge) et NPs de 200nm/graphene/quartz

(bleu).c) Spectres d’extinctions pour des systems NPs/graphene/quartz

(rouge) et NPs/quartz (noir). Un décalage vers les basses longueur

d’onde de 13nm est détecté pour les NPs de 140nm et de 170nm. . . . 159

9


7.15 Spectres Raman montrant l’évolution de a) la bande G et b) la

bande 2D du graphene pour des interface graphene/quartz (noir), NPs

de 140nm/graphene/quartz (rouge), NPs de 170nm/graphene/quartz

(bleu) NPs de 200nm/graphene/quartz (mauve) .c) Spectres d’extinctions

pour des systems NPs/graphene/quartz (rouge) et NPs/quartz (noir). 160

7.16 a) Spectres d’extinction pour des NPs de 200nm de diamètre et de

300nm de périodicité déposées sur un substrat verre/film d’Au (50nm)/couche

en SiO2 (6nm) en fonction de l’indice de réfraction du mileiu envi-

ronnant. b) Décalage du mode LSP à basse longueur d’onde (520nm)

en fonction de l’indice de réfraction du milieu environnant . . . . . . 161

7.17 Suivi de l’exaltation de la bande G du graphène pour des sytèmes

couplés film d’Au/graphène/NPs d’Au en fonction du diamètre des

NPs métalliques. LA valeur maximale de ce facteur est atteinte pour

les NPs de 140nm et vaut 460. . . . . . . . . . . . . . . . . . . . . . . 161

10

List of Tables

3.1 Comparison of the numerical and experimental results of all the ob-

served plasmonic modes for nanoparticles having a diameter of 200

nm and a 300 nm periodicity, for excitation in normal incidence. . . . 79

4.1 Comparison between the expected and the measured values of the

SiO2 spacer layer thickness. . . . . . . . . . . . . . . . . . . . . . . . 92

4.2 Absorption percentage for NPs of different diameters on 100 nm

SiO2/50 nm au film substrate. . . . . . . . . . . . . . . . . . . . . . . 103

6.1 Calculated enhancement factors of G-band and 2D-band for NPs of

different diameters deposited on monolayer graphene on an Au film. . 134

6.2 Calculated enhancement factors of G-band and 2D-band for NPs of

different diameters deposited on binolayer graphene on an Au film . . 136

6.3 Calculated enhancement factors for the NPs of different diameters for

both the G-band and the 2D-band. . . . . . . . . . . . . . . . . . . . 136

11

General Introduction

“There is plenty of room at the bottom” [13] was the early seed of nanotechnol-

ogy planted by Feyman in an after dinner talk in 1959. He believed that much more

could be done if we considered the possibility of direct manipulation of atoms. At

the time, his idea did not recieve a positive response from the scientific community.

Ten years after in the 1970s, the world nanotechnology was more commonly used

and the field started to emerge [14, 15]. Shortly after, the interest in the nano-

world fastly increased, and remarkable advances were reported. One of the research

fields that branched from nanotechnology is plasmonics which is the study of optical

phenomena related to the electromagnetic response of metals [16]. The possibility

to manipulate light at the subwavelength scale using metallic nanostructures al-

lows the introduction of fascinating applications. The promises of plasmonics [17]

explain the current boom in this research topic. Plasmonics has a high potential

for future applications in new superfast computer chips [17], cancer treatment [18],

ultra-sensitive molecular detectors [19, 20], invisible material with negative refrac-

tive index [21, 22, 23], data storage [24], optical data processing [25, 26], quantum

optics [27], and optoelectronics [28].

The basic forms of plasmonic resonances are delocalized plasmonic modes or

surface plasmon polaritons (SPP) and localized surface plasmon resonances (LSPR).

The first are surface waves with a propagative nature confined near metal/dielectric

interfaces [29] and the second is localized to nanoparticles and associated with a

high local electric field enhancement. Metals and especially gold (Au) and silver

(Ag) are most commonly used as plasmonic building blocks. Extensive research

showed the potential of both SPR and LSPR in various applications, however much

less attention was given to the coupling of these two modes.Plasmonic interfaces

with metallic NPs deposited on top of metallic thin films give the possibility to

excite both localized and delocalized modes, where possible overlap, interaction

and coupling between the two arise. Several separate studies investigated coupled

NP/film systems either with the gold NPs deposited directly on the metallic films

[30], or with a spacer layer, commonly dielectric [31, 32, 33, 34].

Metasurfaces, defined as artificial sheets material with sub-wavelength thick-

ness and electromagnetic properties designed on demand [35]are highly appealing

for optical applications. Coupled NP/film systems and MIM interfaces allow high

tunabilty of the reflectance, absorption, and transmission and thus consist a highly

promising plasmonic metasurface.

Even with all the work already done, a comprehensive understanding of the

13

Chapter 0 List of Tables

underlying physics of these systems still lacks. This motivated us to perform a full

systematic study on coupled NP/film systems in an attempt to better understand

the optical phenomena behind them, and to optimize the parameters for future ap-

plications on similar systems. In this thesis work, we have studied coupled NP/film

systems with different spacers. We investigated metal insulator metal structures

with ultra thin dielectric spacer layer, we also studied the influence of a systematic

increase in the dielectric spacer layer on the optical properties. On the other hand

the potential of using novel material such as graphene as a subnanometer spacer layer

was also investigated. A brief summary on the different chapters of this manuscript

will be introduced in the following subsection.

The first chapter of this PhD dissertation revisits briefly the theory behind

plasmonics. The model describing the interaction of electromagnetic waves with

metals is explained. And the origin of Bulk plasmons as well as localized and

delocalized plasmons is discussed, along with the basic formulas governing these

different phenomena. After which, an overview of the different plasmonic modes

associated with both simple and complex nanostructures are introduced. The second

part of this chapter discusses coupled NP/film systems which is at the heart of

this thesis work. With a narration on the different studies on this topic achieved

worldwide by other groups.

In chapter two, we introduce the different experimental methods and techniques

involved through out this thesis. The aim of this chapter is to facilitate the under-

standing of both the fabrication and the characterization techniques which were

used. Fabrication techniques as thin film deposition and preparation of NPs by

electron beam lithography (EBL) are explained in details. The different spectrom-

eters and microscopes we used to acquire the experimental data are also described.

We believe explaining all the technical details in a separate part at the beginning of

the thesis should favor the clarity and sequence of the different experimental results

that will be introduced later.

In chapter 3, we show both the experimental and numerical results which were

performed on MIM interfaces with thin spacers and Au NPs deposited on top by

EBL. The example of a MIM interface with a 6 nm SiO2 spacer is discussed in detail,

since it exhibits a rich plasmonic interface. Investigation of the different plasmonic

modes is performed, with the identification of the localized and delocalized modes

using electric field maps. We also observe a Fano resonance in a broad wavelength

window of 650 to 800 nm. The origin of this Fano resonance is further investigated

and an analytical model is developed to better explain the underlying interactions

resulting in this asymetric lineshape. Further angle resolved measurements are per-

formed, which evidence the tunability of the different plasmonic resonances and

further validates our understanding of their nature.

In chapter 4, the study of MIM interfaces is continued with an emphasis on the

importance of dielectric layer thickness. We perform a systematic study with eight

samples exhibiting gradually increased dieletric spacer thickness. The evolution of

14

List of Tables

the different plasmonic modes versus the increase in the thickness is investigated.

A cross comparison with numerical results is performed which is in good agreement

for small NPs, but shows larger discrepancies for big NPs. The role of the numerical

aperture is exploited, in a successful attempt to explain the previously mentioned

discrepancies. Overall this chapter provides a data bank of extinction spectra with

plasmonic resonances covering most wavelengths for the different thicknesses.

In chapter 5, we attempt to use a somehow unconventional material “Graphene”

as a subnanometrer spacer. Graphene being a non-dielectric material with extraor-

dinary electronic and optical properties shows an unexpected behavior compared

to the dielectric spacer case. The presence of graphene induces a blue shift and

a sharpening of the plasmonic resonance, which is hard to explain based solely on

optics. Further investigations with simplified samples where Au and Ag NPs are

fabricated directly on graphene coated glass substrates without Au film, confirms

that this behavior results from graphene doping when in contact with metals. Such

a finding is highly interesting since it sets the path for graphene based optoelectronic

devices.

In the last chapter, we discuss the possible application and advantages of cou-

pled NP/film systems. We perform RI sensing on different interfaces and we show

that MIM interfaces are the most promising with a figure of merit (FoM) as high

as 4.2 with more than a 200% increase compared to similar NPs designed directly

on glass. This FoM is one of the highest reported in the literature in this wave-

length range. We proceed by performing surface enhanced Raman spectroscopy

(SERS) on graphene-based interfaces. We show that the coupling between NPs and

metallic films leads to a high localization of the electric field, resulting in remark-

able enhancement factors in graphene Raman bands compared to interfaces without

metallic films.

We conclude our work with a brief summary on the main results achieved in

this thesis. We also discuss the perspectives and future work that still to be done,

especially that a lot of our results opened doors for new research directions.

15

1 Glamour of plasmonics

1.1 Introduction

Long before science investigated plasmonics, artists mastered this optical phe-

nomena by mixing glass with gold powder for producing beautiful art. Stained glass

windows in churches and the Lycurgus cup which date back to the 4th century AD

were the beautiful results of this optical phenomenon. The Lycurgus cup gained its

fame because of its changing color seen as red or green depending whether it is illu-

minated from the outside or inside, which back then seemed magical. However the

physics behind these charming colors is even more beautiful and its understanding

started with Gustav Mie in 1908. This chapter introduces the basic theory on the in-

teraction of metals with the electromagnetic (EM) waves at optical frequencies as an

introduction to Plasmonics. Fundamental equations governing the behavior of delo-

calized surface propagating polaritons (SPP) on metallic films and localized surface

plasmon resonances (LSPR) on metallic nanoparticles are reviewed. An overview

of the different plasmonic modes observed since the early rise of plasmonics is also

presented. Finally the importance of metallic NPs/Metallic film interfaces with and

without a spacer layer is discussed. The state of art concerning these hybrid systems

is presented with a short summary of the milestones achieved in the field and an

emphasis on the principal applications developed.

1.2 Plasmonics going down to the nanoscale

Interaction between EM field and metals are governed in the classical limit by

Maxwell’s equations. For a perfect conductor, the propagation of the EM field is

not possible and the net electric field is always assumed to be zero. Indeed the

free conduction electrons deflect when an EM field is applied to cancel out the field

inside the conductor. For low frequencies, metals are highly reflective and can be

considered as perfect conductors with no electromagnetic field propagating through

them. However as frequency increases towards the near infrared and the visible

range, the oscillation of the external field becomes comparable to the characteristic

relaxation time of metals, thus the electrons can no longer stay in phase. As the

frequency increases, the EM field penetration increases where further increase of

the frequency towards the ultraviolet changes the properties of the metal from a

perfect conductor to a dielectric allowing the propagation of EM waves. Also in this

17

Chapter 1 Glamour of plasmonics

regime, transitions between electronic bands in Noble metals occur leading to high

absorption of light. The high dependence of the optical response on the frequency

is described by a complex dielectric function ε(w). The threshold frequency for the

propagation of the electromagnetic waves in metals is called the plasma frequency.

Above this frequency, the collective motion of free electrons can be considered as

an electron gas with a plasma oscillation driven by an external incident electric

field oscillator. Metal insulator interfaces modify this plasma oscillation leading to

surface plasmon polaritons (SPP) in two dimensions. Similarly, the oscillation of the

free electrons present in small NPs results in localized surface plasmon resonances

(LSPR).

1.2.1 Bulk Plasmons

The Plasma model is commonly used as a simple classical model to explain the

interaction between EM waves and metals. A gas of free electrons of mass m and

density N is considered freely propagating against a fixed background of positively

charged nuclei. This simplified approach neglects the electron-electron interaction

and the lattice potential. An applied external electric field E(t) = E0 exp(−iwt)causes the electrons to oscillate, these oscillations are damped by collisions with a

characteristic frequency γ = 1/τ , where τ is the relaxation time of the free electron

gas (t is in the order of 10−14 s at room temperature). For this classical approach,

the equations of motion for an electron under the effect of an electrical field E is:

mx +mγx = −eE(t) (1.1)

which has as solution x(t) = x0e−iwt, where x0 is the complex amplitude that shows

the phase shift between the driving field and the response. The solution can also be

expressed as:

x(t) =e

m(w2 + iγw)E(t) (1.2)

The displaced electrons creates a polarization P = −Nex, where N is the electron

density and e the charge of one electron. With this new polarization the dielectric

displacement can be written as:

D = ε0E + P = ε0εE (1.3)

= ε0

(1 − w2

p

w2 + iγw

)E (1.4)

and the dielectric constant becomes:

ε(w) = 1 − w2p

w2 + iγw(1.5)

18


with wp defined as the plasma frequency for the collective oscillation of the electron

plasma and equal to wp =√Ne2/ε0m. At large frequencies in the limit wτ ≫ 1,

the imaginary part of eq (1.5) can be neglected and the dielectric constant ε(w) can

be expressed as:

ε(w) = 1 − w2p

w2(1.6)

For transverse waves the dispersion relation for electromagnetic fields can be deter-

mined from k2 = εw2/c2 , where k is the wavenumber, and c the velocity of light.

Eq (1.6) becomes:

w(k) =

√

wp2 +k2

c2(1.7)

At this point we can define two regimes: the first for w < wp where no propa-

gation of the electromagnetic field below this plasma frequency is possible. And the

regime for w > wp where the propagation occurs with a group velocity vg = dw/dk.

The quanta of the electrons collective oscillation is defined as bulk plasmon. Bulk

plasmons are longitudinal waves for this reason coupling to the transverse electro-

magnetic fields cannot be achieved, which explains why they cannot be excited under

direct illumination. These features can be seen in Fig. 1.1 which shows the disper-

sion relation of the free electron gas. It is important to note that for noble metals

near the interband transitions the limitation of this model increases and it can no

longer describe the complete picture.

Figure 1.1: Dispersion relation of the free electron gas and the dispersion of light[1]

19


1.2.2 Plasmon resonance in metallic film

Plasmons are the result of free electrons oscillation by an external excitation.

For this reason metals have been most widely used for plasmonic systems especially

Au and Ag. For a dielectric/conductor interface, the electromagnetic excitation

of delocalized plasmonic modes called surface plasmons polaritons (SPPs) can be

achieved (see Fig. 1.2). These propagating surface waves are excited through the

Figure 1.2: Surface plasmon polariton at gold/air interface

coupling of the incoming electromagnetic field with the oscillations of the electron

plasma in the conductor. The coupling is only possible when the wave vector of

the incoming light matches that of SPP at the same angular frequency. Solving the

Maxwell’s equations for the metal/dielectric interface with the continuity conditions

at the boundaries yields the dispersion relation (β) of the SPP propagating at the

surface, for a full derivation refer to Plasmonics: Fundamentals and Applications by

Maier[2]):

β = k0

√ε1ε2

ε1 + ε2

(1.8)

where k0 = w/c is the vacuum wave vector of light with frequency w, ε1(w) the

dielectric function of the metal and ε2 the real dielectric constant of the dielectric

medium. Plugging eq (1.6) for the frequency dependent dielectric function ε1(w) of

the metal in eq (1.8) results in a non-linear dispersion relation w(kSP P ) for SPP.

Comparing it with the linear dispersion relation of white light traveling in the di-

electric media as shown in Fig. 1.3, one can see that no matching between the wave

vectors at the same angular frequency is possible, since the dispersion of the SPP

falls below the light cone. For this reason, direct excitation of SPPs by light is not

possible and specific schemes are required to overcome this mismatch. Several ex-

citation schemes have been developed such as the prism coupling, grating coupling,

excitation using highly focused optical beams, near field excitation, and coupling to

integrated photonic elements [2]. The main excitation schemes specific to this work

are briefly discussed.

20


Figure 1.3: Dispersion relation of surface plasmons compared to light in vacuum

and in the dielectric medium[1]

1.2.2.1 Prism coupling

Phase matching of the SPPs can be achieved through a 3 layer system where the

metal film is sandwiched between two insulators. The first insulator is commonly

considered as air with ε = 1. If the incident beam on the insulator with higher index

of refraction commonly a prism undergoes total internal reflection it will form an

evanescent electromagnetic field. When the in-plane component of the impinging

photon wavevector coincides with air-metal SPP wavevector, resonant light with a

propagation constant β = wc

√εprism sin θ tunnels through the metal film with an

in plane momentum sufficient to couple with the SPPs at the interface between

the metal and the lower index dielectric i.e metal/air interface. The most common

configuration of the prism coupling is the Kretchman configuration, where the metal

film is on top of the prism as shown in Fig. 1.4. Illumination through the prism

Figure 1.4: excitation of SPP through prism coupling, the kretchman configuration

with angles higher than the total internal reflection tunnels through the metal film

allowing the excitation of SPPs at the metal/air interface. sec. 1.2.3 shows the

dispersion relation for the SPP through prism coupling. The phase matching allows

excitation of SPPs at the metal/air interface, however no excitation of SPPs at the

metal/prism interface is possible [36].

21


Figure 1.5: Prism coupling an SPP dispersion[2]

1.2.2.2 Grating coupling

Another commonly used scheme to overcome the mismatch in wavevector be-

tween the in-plane momentum kx = k sin θ of light and β of the SPPs is achieved

by diffraction effects [2]. Diffraction can be realized by patterning the metal surface

with a grating. For a one dimensional grating the phase matching is achieved when

β = k sin θ ± nΛ (1.9)

where Λ = 2π/a, θ the incident angle, and a the lattice constant as shown in Fig. 1.6

Figure 1.6: Grating coupling for light with a wavevector k incident on a metal

grating surface with a periodicity a

22


1.2.3 Metallic nanoparticles and their plasmonic properties

The second form of plasmonic excitation are the localized surface plasmon (LSP)

modes. On contrary to the SPPs, these are non propagating excitation of the elec-

tron charge oscillation of metallic NPs. They result from the scattering of a sub-

wavelength metallic NP in an oscillating EM field. The particle exerts a restoring

force on the driven electrons which arises a resonant mode that amplifies the field

both inside and in the near field outside the particle (see Fig. 1.7). LSP resonances

are transverse and thus can be excited directly through an incident EM wave. Silver

and gold exhibit an LSP resonance in the visible regime and for this reason, they are

the most famous plasmonic materials. For a better understanding of the interaction

Figure 1.7: localized surface plasmon on a metal nanoparticle in the presence of

an electromagnetic field

of the electromagnetic field with a particle of size d ≪ λ, i.e when the particle is

much smaller than the wavelength of light in the surrounding medium, the simple

quasi-static approximation can be used. In this approximation, the spatial compo-

nent of the electromagnetic wave is constant over the volume of the particle and no

retardation in the field is considered. This allows the calculation of the spatial field

distribution by assuming the simplified case of a particle in an electrostatic field.

The harmonic time dependence can then be added once the solution for the spatial

distribution is calculated. This approximation of the scattering problem is valid

for nanoparticles smaller than 100 nm. The simplest example for demonstrating

the underlying physics is the case of a homogeneous, isotropic metallic sphere of

dielectric permittivity ε and a radius a under a uniform static incident electric field

E = E0z, and surrounded by an isotropic and non absorptive dielectric medium

with a permittivity εm. The electromagnetic fields are parallel to the z direction at

a specific distance from the sphere. This becomes an electrostatic problem, where

the solution of the Laplace equation for the potential ∇2φ = 0 allows us to calculate

23


the electric field E = −∇φ. A detailed derivation can be found in several textbooks

[2, 37]. Calculations show that the applied field will induce a dipole moment inside

the sphere. This dipole moment has the form:

p = 4πε0εma3 ε− εm

ε+ 2εm

E0 (1.10)

From (1.10), the polarizability α is

α = 4πa3 ε− εm

ε+ 2εm

(1.11)

The resonance condition for the polarizability is satisfied when the denominator

| ε + 2εm | is minimum. For small or slowly varying Im [ε] around the resonance,

the condition simplifies to

Re [ε(w)] = −2ε (1.12)

This relation is called the Frohlich condition [36]. The equation describing the

electromagnetic surface modes can be solved by means of the Laplace equation [36]:

εm(w)l + ε(w)(l + 1) = 0 (1.13)

For a Drude like dielectric function without damping εm(w) = 1 − w2p/w

2, and the

embedding medium being vacuum ε(w) = 1 , the eigenfrequencies of the surface

electromagnetic modes are:

wsphl =

√l

2l + 1wp (1.14)

where l is the order of the spherical harmonic in which the deformations of the

electron gas is decomposed due to the spherical symmetry of the problem. The

frequency of the dipolar (l = 1) surface plasmon mode of a sphere in vacuum is

given by:

wsphl=1 =

wp√3

(1.15)

For the mode (l = 1), the particle is considered as a perfect dipole. This resonance

is the most common for small spherical nanoparticles. When the dielectric constant

εm of the surrounding media increases it induces frequency red shifts. The depen-

dence of this resonance on εm is the main reason why metallic nanoparticles are

so appealing for sensing applications. Both the inner and outer electric fields are

indeed related to the polarizability α and and are resonantly enhanced when α is in

resonance. The enhanced electric fields become:

Ein =3εm

ε+ 2εm

E0 (1.16)

24

1.3 In depth view of different plasmonic modes

Eout = E0 +3n (n.p) − p

4πε0εm

1

r3(1.17)

where n is the unit vector in the direction of the point of interest. This field enhance-

ment plays the key role for all research on metal nanoparticles in optical devices.

The enhancement of the polarization or the dipole plasmon resonance induces an

enhancement in the absorbance and scattering (and thus in the extinction which

is the summation of the two) of light by the nanoparticles. The scattering and

absorbance cross sections can be calculated via the Poynting vector[38]:

σsca =k4

6π| α |2= 8π

3k4a6

∣∣∣∣ε− εm

ε+ 2εm

∣∣∣∣2

(1.18)

σabs = kIm [α] = 4πka3Im[ε− εm

ε+ 2εm

](1.19)

From Eqs (1.14) and (1.15) it can be seen that for small particles the absorption

efficiency dominates over the scattering as it scales with a3. So far we discussed

the case were the particles are small (d ≪ λ) and the quasi static approximation

holds. However for bigger particles the electrostatic solution is no longer valid due

to the significant phase changes of the driving field over the particle volume, and a

full electrodynamics approach is necessary. In 1908 the Mie theory was developed

and explained the scattering and absorption origin of electromagnetic radiation by a

sphere[39]. A full description of the Mie theory can be found in several references[38,

39, 40].


In the previous section, we introduced the basics of plasmonic resonances, both

for delocalized resonances on metallic films and for localized modes in small metal

nanoparticles. However the recent progress in fabrication techniques provided a

large variety of innovative NPs. Researchers have indeed reported the fabrication of

nanocylinders, nanorods, nano-stars, nano-cones, nanoshells, nano-rings... Moreover

different combinations of NPs have been used, ordered arrays, dimers, heptamers...

This makes the optical response more complex than the first order dipole model pre-

sented earlier. The principal modes that can be excited for different NP geometries

are listed below:

1) Excitation of quadropoles and higher modes for big NPs

2) Excitation of both a longitudinal and a transverse mode for asymetric NPs

25


3) Hybridization between different modes for more sophisticated geometries results

in new resonances.

4) Neighboring particles in arrays can couple leading to shifts compared to the single

particle resonance; when two or more nanoparticles are brought close enough to each

other a gap mode can form.

As for delocalized modes supported by metallic films, these resonances can also

go beyond the traditional resonance when combination of different metal films are

presented. The most commonly studied are the: insulator-metal-insulator (IMI)

system that holds assymetric SPPs and the metal-insulator-metal (MIM) system

that holds symmetric SPPs.

Finally nanoparticles can be brought in proximity to metallic films leading to cou-

pling and hybridization of modes of different nature. This actually is at the heart

of this manuscript. In this section, we will present and define the nature of different

plasmonic modes.

1.3.1 Localized modes in nanospheres

To understand the optical properties of metal nanospeheres and their corre-

sponding plasmonic resonances, one can use the Mie theory [39] which allows the

prediction of the extinction spectra (absorption + scattering) of spherical NPs of

different diameters. For spheres with diameters 3 ∼ 40 nm, absorption dominates

and a sharp LSP resonance occurs at a wavelength λ varying from 520 to 550 nm.

For spheres with a diameter larger than 40 nm, scattering becomes more significant,

the resonance is broadened and shifts to the red displaying size dependent colors

[2, 39]. In fact for particles smaller than 100 nm, the resonance can be explained by

oscillation of dipoles [2]. However as the size increases beyond that up to few hun-

dreds of nanometers multiple resonance peaks appear in the extinction spectrum.

The multiple peaks are the result of the excitation of higher order LSP resonance,

where l (order of the spherical harmonic) is more than 1. In this case oscillation

of quadrupole (2nd order), hexapole (3rd order) and higher multipoles, as well as

dipoles (l = 1) should also be taken into consideration. Fig. 1.8 shows the charge

distribution on a sphere for dipoles, quadropoles and hexapoles.

1.3.2 Localized modes in nanorods

In a previous section we explained how the plasmonic resonance results from the

oscillation of the NP’s electron cloud when its frequency is matched with that of the

incident electromagnetic wave. This resonance depends on the induced polarizabil-

ity, which itself depends on several factors: shape, size, composition of the material

and the surrounding medium. Nanoparticles with broken symmetry geometries such

as rods and cylinders, exhibit two modes at different wavelengths: the transverse

26


Figure 1.8: Charge distributions of dipole, quadrupole, hexapole surface plasmon

are depicted.

and the longitudinal modes. The first is for the polarization on the short axis and

the second is for polarization on the long axis as shown in Fig. 1.9

The longitudinal resonance occurs at lower energies, and thus at higher wave-

lengths since this polarization can be more easily achieved. An important parameter

for rods and cylinders is the aspect ratio defined as the ratio of the particle’s length

to its width. As this ratio is increased for a fixed diameter, both polarizations are

affected but not equally. Since the longitudinal axis is much more polarizable, it is

more sensitive to changes in the aspect ratio. For gold nanorods, the LSPR wave-

length is thus highly tunable (from 550 nm up to 2000 nm) by adjusting the aspect

ratio [41]. However the TSPR remains almost constant around 510-520 nm.

1.3.3 Coupling between neighboring metallic nanoparticles

The LSP resonance of a single nanoparticle is different from that of an ensemble

of particles. Where shifts due to electromagnetic interactions between the localized

modes may occur. Each small NP can be considered as a dipole, with the array of

dipoles interacting through their near-field as shown in the study of Krenn et al [42].

For ordered arrays with particles of size a and interparticle distance d, the dipolar

approximation is justified when a ≪ d and the particles can then be treated as point

dipoles. Within this assumption, two regimes should be identified. The first one is

for closely spaced particles and the second for larger particle separations. In the first

regime, the near-field interactions dominate with the distance dependency scaling

as d−3. The particle array can thus be described as an array of point dipoles inter-

27


Figure 1.9: Schematic showing the longitudinal and the transverse localized plas-

mon mode of a nanorod

acting via their near-field. The spectral shifts observed for an ensemble of NPs may

therefore be treated in the approximation of interacting point dipoles. To determine

the direction of the shift, the Coulomb forces associated with the polarization of

the particles must be considered. Depending on the polarization direction of the

incident light, the oscillating electrons of each particle in the chain can be either

increased or decreased by the charge distribution of neighboring particles[2]. For a

transverse polarization, the second dipole will respond in phase with the polarization

of the first dipole. The in-phase response happens at lower frequencies compared to

the single dipole resonance and as a result the wavelength resonance of the coupled

mode is red shifted. As for the longitudinal polarization, the second dipole respond

out of phase with the induced field in the first dipole. In this case, higher frequen-

cies are required to drive the new coupled mode and a blue shift in the resonant

wavelength compared to the single dipole occurs[43]. This dipole-dipole interaction

model explains nicely why spectral shifts do occur when the nanoparticles are close

to each other, as well as the direction of the shift. However a qualitative estimation

of those shifts remains out of the scope of such a model. The limitation of this

model arises from the Coulomb interaction between the oscillating electrons of the

NPs when they are strongly coupled. This Coulomb interaction distors the particle

plasmon leading to mixing of higher order modes into the coupled mode [43]. For

this reason the “plasmon hybridization” model was introduced [44, 45] to give a

semi-analytical picture in the quasitatic limit where the metals are treated using

Drude model and neglecting the damping effects. A brief introduction of this model

will be presented in the next section.

For the second regime with larger interparticle spacing the distance dependence

28


Figure 1.10: Schematic of near-field coupling between metallic nanoparticles for

different polarization schemes

becomes 1/d and the far field dipolar coupling dominates influencing both the res-

onance wavelength and spectral width of the plasmonic resonance [46, 47]. The

changes in the spectral width are due to the high dependence of the decay time on

the grating constant. This is mainly because different grating constants can change

the damping from evanescent to radiative as shown in the study of Lamprecht et

al[48]

1.3.4 More complex geometries and the hybridization model

Several groups performed extensive work both experimentally and theoretically

to show the plasmonic resonance for different nanoparticle geometries [49, 50]. How-

ever with the emerging advances in the field, a large variety of nanoparticles have

been fabricated, such as rings, shells, crescent, and other complex structures. More-

over fabricating nanostructures with few nm separation is also now possible. This

makes the prediction of the plasmonic properties rather hard. To better understand

and explain the spectral distribution of the plasmonic modes of complex structures

and coupled nanoparticles, Prodan et al proposed a hybridization model in 2003 [44].

They built their model on a similar concept to that of orbital molecular hybridiza-

tion theory but applied to electromagnetic systems which explains the resonances of

complex nanostructures in terms of the well-known plasmonic modes supported by

the basic entities forming them. For example, the energy of the modes of a metallic

shell can also be derived from classical electromagnetic theory. We consider the

example of a metallic shell to better understand the hybridization theory. A shell

is considered as a spherical cavity coupled to a metallic sphere, where the plasmon

modes of the cavity and the sphere interact. This interaction results in the splitting

of the LSPR into two new modes: the bonding plasmon w−, with lower energy, and

29


the anti-bonding plasmon w+, with higher energy as shown in Fig. 1.11. This model

gives an elegant insight into the physical mechanism of coupled plasmons. However

it is limited to the quasi-static limit with the assumption that there is no damping in

the system. For more complex and larger particles where the retardation and damp-

ing effects are significant, full electrodynamical simulations are required. A detailed

Figure 1.11: Schematic of plasmon hybridization in metallic nanoshells, where ws

is the frequency of the sphere, and wc the frequency of the cavity.

analysis with governing equations can be found in ref [44]. This model is an elegant

and simple method which successfully explains the behavior of complex nanostruc-

tures such as dimers, nanoshells, nanostars or nanomatryuskas. Its importance lies

in the intuitive way of explaining the complex nature of the electromagnetic sur-

face modes of nanostructures. A year later, Nordlander and Prodan elaborated this

model in order to study the hybridization in NPs near metallic surfaces [45].

1.3.5 Symmetric and assymetric delocalized plasmon modes inIMI and MIM structures

SPP modes can propagate on insulator-metal-insulator (IMI) as well as metal-

insulator-metal (MIM) structures. For these structures, two metal/insulator in-

terfaces exist, hence two different SPP modes can be excited. The bound SPPs

propagating at each interface may interact with each others and form coupled SP

modes when the separation between adjacent interfaces is comparable to or smaller

than the penetration depth. For an IMI structure, one SPP mode can propagate

at the upper interface and another mode at the lower interface. For a metal film

thick enough, the two modes do not feel each other. However as the metal film de-

creases the two SPP modes couple resulting in both symmetric and asymetric SPP

30


modes. When the superposition of the Ex field component of the SPP modes at the

two interfaces are in phase, they induce a symmetric SPP mode which exhibits a

symmetric charge distribution as shown in Fig. 1.12 (a). When the superposition is

out of phase, the result is an anti-symmetric SPP mode with anti-symmetric charge

distribution as shown again in Fig. 1.12 (a). The anti-symmetric coupling is a low

energy odd mode while the symmetric coupling is a high energy even mode. One

of the important features of IMI structures was first reported by Fuku et al [51].

In their study, they measured the life time of propagating modes, and showed that

the anti-symmetric mode has a long lifetime and is thus called the long range SPP

mode. After this study, several groups measured the propagation length of this SPP

and reported considerably varying depending on the metal and insulator involved, as

well as the metal film thickness and the wavelength [52, 53, 36]. The values reported

experimentally and theoretically are in the range of hundreds of micrometers for a

15 nm Ag film at a wavelength of 632.8 nm [54, 55, 56]. As a result of those long

propagation lengths, IMI structures were used to realize SPP waveguides made of

metallic nanostructures.

The MIM structure also allows the propagation of two SPP resonances which

can couple to form both symmetric and anisymmetric modes as shown in Fig. 1.12

(b). MIM structures have attracted a great deal of attention, especially case of thin

insulator spacers[57, 58, 59, 60, 61, 62]. Antisymmetric modes in MIM structure also

exhibits a waveguide mode character which remains even for very small dielectric

spacer thicknesses going down to zero. This was shown in several theoretical studies

and analysis [58, 59, 62]. These MIM structures and their antisymmetrical mode

are strong candidates for plasmonic slot waveguide with nanometer-sized dielectric

layers. MIM waveguides with SiO2 spacer layer whose thickness decreased down to

3.3 nm were fabricated and a full treatment of the optical properties both experi-

mentally and theoretically was performed by several groups [59, 62]. Finally one of

the most appealing features of these MIM structures is that the dispersion curves

of symmetric modes start at kx = 0. The fact that the dispersion curve is located

in the radiative region indicates that this mode can couple directly with incident

light. For thin Ag film (≤ 50nm), the symmetric SPP mode can be excited directly

through coupling with the light propagating outside the MIM structure.

1.3.6 Gap modes

In the previous section, we discussed coupling the between nanoparticles when

they are close to each other. For a dimer configuration, the two electron clouds can

oscillate either in phase (bonding state) or out of phase (antibonding state) forming

coupled longitudinal and transverse modes propagating parallel and perpendicular

to the dimer axis, respectively. As two nanoparticles are brought together in close

vicinity, a remarkable and highly interesting feature occurs where a region of ex-

tremely large electric field, referred to as a hot spot is generated. Only when the

longitudinal mode is excited and the nanoparticles are brought closer and closer to

31


Figure 1.12: Symmetric and antisymmetric coupled SPP modes in a) IMI and b)

MIM structures

each other an extremely enhanced electric field, several order of magnitudes higher

than the field in a single NP is induced in the gap between the particles as shown in

Fig. 1.13. This high enhancement of the electromagnetic field in a few nm separation

is referred to as the Gap mode.

Figure 1.13: Longitudinal mode for a dimer with small seperation showing a hot

spot in the gap between the NPs.

It is important to note that when the NP sizes are too small or the separation

between particles goes down to less than 0.5 nm, this problem enters the quantum

regime and a different treatment is required [63, 64, 65]. On the other hand when

a NP is brought in close vicinity to a metallic film, a similar effect can occur in

the gap between the particle and the film. An increase in the electromagnetic

32


field enhancement becomes possible as the distance d between the film and the NP

decreases and the generation of a hot spot occurs as shown in Fig. 1.14.

Figure 1.14: Schematic of the generation of a hot spot in the gap between a metallic

nanoparticle and a metallic film.

Gap modes can also be present in MIM and IMI structures where the high

energy or the hot spots are confined in the dielectric layer. This gap mode is highly

dependent on both the nature and the thickness of the dielectric. The high en-

hancement of the electromagnetic field and the generation of the hot spots seem

very appealing for surface enhanced Raman spectroscopy (SERS) measurements. In

1997 hot spots induced by particle aggregation resulted in enhancement factors as

large as 1012 based on microscopic Raman measurements [66]. Furthermore, this

intense local field in a gap-sized region leads to a strong coupling to whatever is

trapped in the gap. This makes the nanoparticle pairs and the film-particle systems

excellent nanoantennas for coupling light into and out of localized emitters [43].

When the material in the gap changes, spectral shifts in the plasmon resonance

frequency occur. This can be used for sensitive chemical detection. Finally the

resonant wavelength highly depends on the distance separation between the NPs,

and can be used as a “ruler” [67, 68] which allows the measurement of the size of

the gap between either two nanoparticles or a particle/film.

1.3.7 The dark side of plasmonics

The traditional spectroscopy technique for probing plasmonic modes consist in

measuring the extinction or absorption cross section. This allows to measure bond-

ing modes or the so-called “bright” modes. An external plane wave can induce an

oscillating dipole in the nanoparticle. The induced dipole increases the scattering

cross section and decays through radiative damping, and thus can be easily mea-

sured. Another plasmonic mode is the “dark” mode also known as the antibonding

mode. Contrary to the dipolar plasmon mode, it has a net dipolar mode close to

zero and it looses energy only by internal dissipation: it is thus much sharper than

a bright mode. Dark modes are very weakly excited by plane waves. This allowed

33


them to hide from optical spectroscopy for quite a long time. Since dark modes

cannot be measured with typical optical spectroscopy, more advanced techniques

were required, such as cathodoluminescence, photo electron emission, and electron

energy loss spectroscopy (EELS). The latter is the most commonly used, and it is

based on exposing a material with an electron beam having a known range of ki-

netic energies. Due to the interaction with the material, some electrons will undergo

inelastic scattering where they loose part of their energy. By analyzing the amount

Figure 1.15: Full modal spectrum of a silver nanodisk with 200 nm diameter and 30

nm thickness on a 30 nm thick silicon nitride membrane. (a) Experimental (blue)

and simulated (red) EEL spectra from the particle edge (top), center (middle),

and the region around half radial distance (bottom), as indicated in red on the disk

sketches.(b) Simulated surface charge distributions (right) generated by assuming

the electron beam positions marked by the cross. Figure taken from ref [3]

of the energy lost, the origin of the interaction between the material and electrons

can be deduced. Plasmon modes both dark and bright demonstrate a common type

of inelastic scattering of electrons.

One of the most appealing theoretical and experimental work investigating gap

modes was done by Scmidt et al [3]. In their work, they measured using EELS dark

and bright modes in silver nanodisks. The position of the focus of the electron beam

with repsect to the particles was changed and the spectra measured for the different

positions varied noticeably especially for positions where the net dipole moment is

zero. Both experimentally and through simulations, they showed the presence of the

metallic nanosphere (dipolar, quadrupolar, hexa..) modes, as well as modes specific

of the cylindrical shape. These modes can be seen in Fig. 1.15 taken from ref [3]: a

(0,1) dark mode, a (1,1) bright mode, and a (0,1) radial breathing mode showing

the strongest electron coupling.

34


1.3.8 Fano modes

The Lorentzian formula has been considered as the fundemental lineshape of

most resonances. In 1961 Ugo Fano evidenced a new type of resonance, characterized

with a high asymetry compared to the symmetric lorentzian form. He discovered

this resonance while doing quantum mechanical study of the autoionizing states of

atoms[69]. This asymetric resonance has the following functional form:

I �

(qγ + w − w0)2

(w − w0)2 + γ2(1.20)

where w0 is the resonance position while γ presents the resonance width. The newly

introduced parameter here is q or the so-called Fano parameter which describes the

degree of asymetry. The Fano resonance arises from the constructive and destructive

interference of a narrow discrete resonance with a broad spectral line or a continuum

[70]. Since its discovery this resonance has been a feature of quantum systems for

a long time. However such an interference phenomena can also occur in classical

systems. For a better understanding of the mathemetical formalism suggested by

Ugo Fano, Fig. 1.16 shows an illustration of eq (1.20) as a superposition of the

lorentzian lineshape of the discrete level with a flat continuous background.

Figure 1.16: Illustration of the Fano profile. Figure taken from ref[4].

According to this assumption, the resonance profile of the scattering cross sec-

tion is:

σ =(ǫ+ q)2

ǫ2 + 1(1.21)

where q is a phenomenological shape parameter which defines the degree of asymetry,

ǫ is the reduced energy defined by 2(E − EF )/Γ . EF is the resonant energy while

Γ is the width of the autoionized state. When | q |≫ 1, the interference is domi-

nated by the discrete resonance which results in a lorentzian peak on top of a weak

background. For | q |= 0 the interference results in a spectral dip and an anti-

resonance in the continuum, finally for q = ±1 the degree of asymetry is maximum

35


Figure 1.17: Normalized Fano profiles with the prefactor 1/(1 + q2)for different

values of q. Figure taken from[4].

[4]. Fig. 1.17 shows the normalized fano profile for different values of the asymetry

parameter q.

We discussed earlier how the plasmon character can be modified through cou-

pling and hybridization between two NPs due to the Coulomb interaction between

their oscillating charge densities. Another way which can induce plasmon charac-

ter modification comes from interference. If multiple plasmonic resonances can be

excited simultaneously in metallic NPs and if these resonances spectrally overlap,

the overall response can be very different from the individual resonance. In the

case where one of these two resonances is a broad bright mode and the second a

discrete sharp resonance, the conditions for a Fano profile are met and the overall

response exhibit an asymetric line shape. Interest in Fano resonances for plasmonic

systems arised for several reasons, mainly since these resonances with their sharp

width and high tunability showed good promises for chemical sensing. Engineering

of different plasmonic structures whose resonance corresponds to a Fano lineshape

was done by several groups. The key parameter is to engineer structures with both a

discrete subradiant or dark mode and a broad super-radiant one. Several geometries

were investigated where the excitation of both of these two modes simultaneously is

possible. The first report of a fano resonance in plasmonics was for a dolmen-type

slab arrangement [71]. Other nanostructures can also have a Fano resonance, such

as the non-concentric ring/disk cavity [72, 73] shown in Fig. 1.18 (a and b). This

structure holds two dipolar modes: one for the ring and another for the disk. The

hybridization of these two modes results in two resonances:

- A low energy dipolar antibonding resonance. The dipole moments of the ring and

disk are aligned oppositely and thus the radiative damping is highly reduced. This

mode can therefore be considered as a dark mode.

- A high energy bonding mode where the dipole moments of the individual ring and

36


disk are aligned and oscillate in phase to form a super-radiant broad mode.

The concentric ring/disk geometry breaks the symmetry which allows an overlap

between the super-radiant dipolar continuum and the narrow quadrupolar ring/disk

cavity. This interference between the discrete mode and the continuum results in

a Fano resonance as seen in Fig. 1.18 (a and b). It is also important to note that

for these structures, changes in the incident angle and in the polarization can affect

the Fano profile and the degree of asymetry since the interference conditions are

changed. Fig. 1.18 (c) shows the change in the line shape versus the change in the

incident angle, it can be observed that the degree of asymetry and the lineshape are

sensitive to the changes in the angle [72].

Figure 1.18: Calculated extinction spectra of non-concentric ring/disk cavity of

partial a) and complete b) filling of the cavity (yellow areas in the inset) with

a dielectric medium of permittivity 1 (black solid), 1.5 (blue dashed) and 3 (red

dotted). The structure is placed on a glass substrate modelled using a permittivity

of 2.04, c) extinction spectra of thenon-concentric ring/disk cavity as a function of

angle of incidence θ defined in the inset. The charge amplitude plots in the panels

to the right show the symmetry of the higher multipolar modes at their resonance

wavelengths. Figure taken from ref [5].

The third family of nanostructures with sharp Fano profiles are the symmetric

NP clusters such as hexamers and heptamers [74, 6]. For such symmetric structures,

the central particle plasmon interacts with the collective plasmon modes of the six

particle ring structure to form hybridized states, Fig. 1.19 (a). Due to the dipole

37


orientation each of the six surrounding particles, the net dipole of the particles

outside is almost equal to the dipole of the single particle. When these two dipoles

interact, they can result in a bonding mode where the dipoles oscillate in phase

and the mode is super-radiant. When the dipoles oscillate out of phase and the

net dipole moment drops down to almost zero (see Fig. 1.19 a), the result is an

antibonding mode. Due to the symmetry of these structures, they can only be

tuned by changing the geometrical parameters such as the distances between the

central and outer particles, or between the outer particles themselves [7]. The first

two parts of Fig. 1.19 (b) shows the effect of changing the geometrical parameters

[6], while the third part shows how the removal of the central nanoparticle can alter

the Fano shape since the coupling between the outer nanoparticles and the inner

one no longer exists, and thus no interference can occur [7].

Figure 1.19: a) Calculated dipole amplitudes of the bonding and antibonding col-

lective dipolar plasmon modes in a gold nanoshell heptamer [6], b) Transmission

spectra showing the effects of coupling in lithographically fabricated gold nanodisk

heptamers. Figure taken from ref [7].

Besides the introduced nanostructures for which a Fano resonance can be ob-

served, it is possible to excite Fano resonances in plasmonic systems in other few

different ways as reported by several groups. Zhang et al [75] showed the possibility

to have a substrate induced Fano resonance where a semi infinite dielectric slab can

introduce coupling and hybridization of metallic nanostructures. For instance in

the case of a nanocube, the nearby dielectic mediates an interaction between bright

dipolar and dark quadrupolar modes. When these two modes interact, they result

in bonding and antibonding hybridized modes. Interference between these resultant

modes thus obeys the condition of a Fano resonance and an asymetric resonance can

be observed [75]. As for Lodwejiks et al [76], they showed that for systems with Au

NPs on top of Au films with a thin dielectric spacer layer, a crossing and anticrossing

between SPP and LSP modes can be produced, which results in asymetric profiles.

They also showed how these Fano resonances are very promising in refractive index

phase sensitivity measurements [76]. The final work which will be discussed here is

38

1.4 Optical properties of film-nanoparticles coupled systems

that of Mikael et al [8] who investigated the possibility of inducing a Fano resonance

via the interference between the sharp LSP resonance of Au nanodisks with a “con-

tinuum” white light resulting from interface reflection as shown in the schematic in

Fig. 1.20 (a). They also showed the high dependence of this Fano resonance on the

incident angle. Changing the incidence angle can break the asymmetrical profile.

They attributed that to changes in both the amplitude and phase of the LSP mode

and the continuum which can modify the interference condition (see Fig. 1.20 b).

Figure 1.20: a) Schematic of interface reflections and a summary of the Fano res-

onance interference conditions, b) Specular reflection spectra for different angle of

incidence. The assymetric line-shapes is for angles close to the critical angle of

the glass-air interface, θc = 41.1° . Dotted lines are experimental data while the

solid lines are fits to the Fano formula. Figure taken from ref [8].

1.4 Optical properties of film-nanoparticles coupled

systems

Surface plasmon polariton (SPP) and localized surface plasmon (LSP) have at-

tracted numerous researchers due to their high technological potential. As previously

explained SPP’s are surface waves confined near a metal dielectric interface that can

propagate over large distances, making them appealing for applications in biosensing

[77, 29]. On the other hand, LSP resonance is a localized resonance condition that

massively enhances the electromagnetic field in the vicinity of a metal nanoparticle

when the NP have dimensions much smaller than the excitation wavelength. LSP

resonance is very sensitive to changes in the NP’s dimensions, the dielectric constant

of the surrounding media and the nature of the substrate. Because of intense local

electrical field enhancements and sharp resonance excitation peaks, metallic NPs are

39


of great interest for applications in surface enhanced Raman spectroscopy (SERS)

[78], chemical and biological sensors [79], cancer treatment [80] and light harvesting

[81, 82, 83]. Recently, the plasmonics community have witnessed increased interest

in the potential of SPP and LSP combinations by investigating metallic NPs on

top of metallic thin films both in the presence and absence of spacer layer. These

systems seem to be very appealing since previous work revealed that they exhibit en-

hanced optical properties and larger tunability of their spectral properties compared

to uncoupled systems [45, 84, 31, 85, 30, 32, 33, 86, 87, 34, 88, 89, 90]. A complete

comprehensive understanding of these systems is far from being achieved due to the

numerous parameters controlling their properties. Of which the most important are

the changes in the nature and thickness of the spacer layer sandwiched between metal

NPs and metal films, changes in the metals used to fabricate the films and particles,

as well as the geometry of NPs and their distribution whether arrayed or random

which can vastly influence the optical properties and the plasmonic resonances. In

the last few years, several studies by many groups were conducted, putting this

topic on the right track. It is crucial to formulate an overview on the early works

on coupled systems and the progress achieved ever since [91]. The following section

will be dedicated to this brief overview.

1.4.1 Breakthroughs in the physical understanding of coupled

systems

In 1983, the first study of coupled metal NP-metal film system was conducted

by Holland and Hall [92]. In this early study, they measured reflectivity spectra in an

Attenuated Total Reflectance configuration (ATR) of Au and Ag islands deposited

on a silver film with LiF spacer of different thicknesses. Their first step was to

measure the SPP of the silver film before the deposition of the spacer layer and

the metallic islands. Those measurements were then repeated after the deposition

of the spacer layer and the metal islands which resulted in an angular shift in the

minimum in reflectance. Those results confirmed for the first time the shift of the

SPP mode due to the coupling with the NPs via the spacer layer. They also changed

the incident wavelength and reported the dependency of the angular shift on the

incident wavelength. The LSP resonance wavelength of the metal islands induced

the biggest shift in the reflectivity measurements. A year later, Holland and Hall

performed further measurements [93] on the the Au and Ag islands on silver film. In

their new set of experiments, they measured the reflectance with normal illumination

on the metal island side. This illumination scheme did not allow SPP excitation since

there was no matching of the SPP dispersion with that of light. For Ag islands,

they reported a single minimum in reflectance corresponding to the LSP resonance.

Comparing the LSP resonances for spacers of different thicknesses on Ag film to

that of the LSP of Ag islands without a film, they noticed that for small thicknesses

(< 50 nm) the resonance red shifts and for larger spacer thicknesses (beyond 50

nm) the resonance blue shifts. The shift of the LSP resonance once again confirmed

40


the coupling induced by the metallic NPs-metallic film configuration. The more

interesting studied case was that of Au islands, where reflectivity measurements

showed two dips: one at 350 nm and another at 540 nm. The 540 nm resonance

corresponded to the Au NP LSP, and the 350 nm resonance corresponded to the

excitation of the film SPP. The phase matching which allowed the excitation of SPP

was achieved through the evanescent waves with high k vector. This excitation of

SPP is possible for spacer layer of low thickness below the penetration depth of

the evanescent field. However this SPP excitation is less efficient compared to the

grating excitation due to the broad spectrum of k vectors. Almost ten years later

Kume et al [94] further confirmed the coupling between the SPP and the NPs. In

their experiment, they excited the SPP through angular phase matching and the

evanescent field of the SPP then excited the LSP of the NPs. A year later, Stuart

and Hall [95] investigated a different scheme of interaction between metal NPs and

metal films. In their case, the illumination occured from the particle side. They

showed that the particles excited the delocalized modes and the delocalized modes

further excited the neighboring particles. They also studied the effect of the spacer

layer thickness. In conclusion their work confirmed that the interaction between

SPP and metallic particles may enhance the absorption.

1.4.2 Coupling between ordered arrays of metallic nanoparticlesand metallic film

Early studies confirmed the possibility of coupling between SPPs and LSPs in

hybrid NP-film systems, with recorded shifts in the resonances of both localized and

delocalized modes compared to non hybrid systems as well as enhancement in the

total absorption. Two main schemes were investigated:

- The first is when phase matching is used to excite SPP whose evanescent field can

then excite localized plasmonic modes on the NPs.

- The second when the illumination is from the NPs side where the evanescent field

of the localized mode excites the SPP on the metal dielectric interface.

The wavelengths dependence on both SPP and LSP show high potential of tun-

ability for similar systems. On the other hand, enhancement in the absorption and

electromagnetic field induced from the interaction between localized and propagat-

ing systems is very appealing for optical applications especially SERS, sensing, and

optical absorbers. A main disadvantage of these coupling schemes remains the low

efficiency due to the broad spectrum of k vectors in random nanoparticles.

Motivated by the potential of coupled systems and to overcome the low effi-

ciency problem, increased works were reported on ordered arrays of metal NPs on

metal film systems both in the presence and absence of dielectric spacer layers. Re-

cently advances in fabrication techniques such as electron beam lithography (EBL)

and the increase in computational power allowed further investigation and under-

standing of the underlying physics and the coupling mechanisms of similar plasmonic

41


systems. The first study on arrays of Au NPs on Au film without a spacer layer was

reported in 2004 by Félidj et al [96]. In their study the presence of the gold film

induced two plasmonic modes in the extinction spectra: a low wavelength mode at

525 nm associated with the isolated LSP resonance, and another mode at higher

wavelengths around 650 nm resulting from the collective LSP resonance. Collective

LSP resonances are common for aggregated colloids where the separation between

the particles is very small compared to the interparticle distance studied by Félidj

and his coworkers. However they believed that the observation of this LSP resonance

was achieved with the help of the gold film which provided a suitable channel for the

propagation of the evanescent field allowing the interaction between nanoparticles

with separations beyond the typical separation length for coupled LSP thus demon-

strating a new interesting property of these systems. Other groups have showed

the possibility to excite on similar samples both a low wavelength (525 nm) vertical

dipolar LSP resonance and a grating coupled SPP at higher wavelengths [30]. The

response of the two modes to the variation of the interparticle distance confirmed

their nature. Localized modes are indeed insensitive to changes in the interparticle

distance however SPP are highly sensitive to those changes as indicated in eq (1.8).

Several studies investigating the coupling between localized and delocalized

modes in NP-film systems in the presence of a dielectric spacer layer were also

conducted. Linden et al [97] studied the optical properties of particles fabricated

by EBL with different shapes and sizes on top of an indium tin oxide (ITO) layer.

Once again the interplay and interaction between LSP and SPP were evidenced and

the dependency of SPP on the interparticle was confirmed. Also Cesario et al [33]

investigated structured interfaces made of Au NPs separated from an Au film by a

50 nm ITO spacer. They observed multipeaks as a result of the coupling between

the delocalized mode in the film and the localized mode of the NPs, similar to what

Hohenau and Krenn [30] reported on hybrid systems. Cesario et al, reported a

SPP mode sensitive to the interparticle distance and a second independent mode

associated with the localized modes of NPs. Besides the common observed modes

reported by most groups both in the presence and absence of a spacer layer, other

modes specific to the geometry and spacer thickness of the studied systems were

reported, such as the Bragg induced SPP in thin spacers reported by Felidj et al

[96], or gap modes localized in the spacing between the NPs and the metal film

[98] with high local electromagnetic field enhancement which is very appealing for

SERS measurements. Several theoretical and experimental studies were done later

to investigate the effect of the distance between the nanoparticle and the thin film

on the behavior of localized and delocalized modes [84, 31, 32].

1.4.3 Potential applications in NP-film systems

Possible applications of coupled particle-film systems have been exploited by

many groups, since these coupled systems can be an essential building block for

42


metamaterials design either to perfectly absorb light, or towards perfect transmit-

tance, and also as plasmonic waveguides for biosensing. In 2010, Hao et al [9]

performed numerical simulations using FDTD and observed that plasmonic systems

of Au NPs on top of Au film with an Al2O3 dielectric spacer layer of 10 nm thickness

had the potential of a metamaterial absorber at optical frequencies. Their experi-

mental results showed absorption up to 88% around 1.58 µm as shown in Fig. 1.21.

According to Hao et al this absorption is due to the excitation of electric and mag-

netic dipoles which conducts to light trapping and its dissipation by ohmic losses.

Figure 1.21: a) Geometry of the sample studied by Hao et al b) SEM images of

the NPs c) Measured and simulated absorbtion spectrafor Wx=170 nm, Wy=230

nm, t =40 nm, d= 10nm, h=50nm, and a = 30 nm at 20° incidence angle. Figure

taken from ref[9].

Also A.Moreau et al [10] studied gap modes between Ag cubes and Au film with

an optimized dielectric spacer between 5 and 10 nm. They showed that by focusing

on the gap modes between the silver cubes and the gold film, they could fabricate

controlled reflectance surfaces as shown in Fig. 1.22. The advantage of their system

was in the random deposition of the silver cubes with low surface coverage about

17% resulting in a low reflectance of 7% (see Fig. 1.22). The random deposition

technique makes it appealing for future commercial applications.

Also M. G. Nielsen et al [11] showed that the fabrication of optical absorbers

over a broad range of optical wavelengths could be achieved through Gap Plasmon

Resonators (GPR). Their structure consisted of a 20 nm SiO2 dielectric spacer layer

sandwiched between Au film and Au NPs. They achieved a 94% of absorbtion over a

wavelength band of 400-750 nm by fabricating continous layers of GPR with different

sizes as shown in Fig. 1.23 taken from ref [11].

Further studies to both increase the efficiency of the absorption and the wave-

length range, with a commercial production cost were pursued by several other

groups. Elbahri et al [99] demonstrated a remarkable enhancement with a lithog-

raphy free process reaching almost 100% absorbency over wavelengths from the

ultra-violet to the near-infrared. An additional advantage of their samples is the

possibility to coat it on flexible polymers. Other groups as M. Yan et al [100] focused

on mass production and thus developed an absorber of metal NPs-dielectric-metal

43


Figure 1.22: (a) Geometry of the sample studied, Au film with Ag nanocubes sep-

arated by a 10nm polymer spacer layer (n=1.54) (b) SEM image of the fabricated

optical metamaterial absorber. (c) Experimental reflectance for normal incidence,

normalized to the gold film, for surface coverages of 7.3% (thin solid line) and

17.1% (thick solid line), compared with numerical simulations of uniform cubes

(4.2% surface coverage, dotted line) and a model including size dispersion. Figure

taken from ref [10].

film where the NPs were formed by annealing thin Au films. Beyond the high ab-

sorption, coupled NP/film systems can be designed to yield transparent conducting

metals. ITO is the most common transparent metal, however in a recent study

Elbahri et al [101], showed that through depositing a polymer with Ag nanocubes

on a 25 nm Au film coated glass, the transmission is 80% which is comparable to

that of ITO but with the advantage of remarkably higher conductivity. Also Z.-Q.

Liu [102] showed through numerical simulations that by depositing nanostructures

on the two sides of the metallic film the double sided metal NPs/metal film/metal

NP could achieve a near unity transparency.

On the other hand, highly efficient plasmonic waveguides, optical sensors, and

increased electromagnetic field enhancements are considered as a universal quest

and a common application for most plasmonic and photonic systems. Studies on

coupled NP/film systems by many groups showed very appealing potentials for such

applications. D. R. Smith and coworkers [103] showed that the fabrication of Ag

nanocubes on Au film leads to the formation of waveguide cavity like modes with

highly interesting optical properties, especially the reflectance tunability and chem-

ical sensing. In another work Cui et al, put into evidence the advantage of using the

44

1.5 Conclusion

Figure 1.23: (a) Geometry of the sample, the unit cell is quadratic of size ▲ having

four Au NPs of different sizes. (b) Schematic representation of the layered sample

structure. (c) SEM image of the fabricated GPRs with particle diameters d1 =

60nm, d2 = 80nm, d3 = 100nm and d4 = 120nm and unit-cell size ▲ = 340nm.

(d) Reflection spectra for unpolarized (solid red curve), x-polarized (dashed blue

curve) and y-polarized (black dash-dotted curve) light. Figure taken from ref [11].

coupled SPP and LSP resonances in hybrid systems in order to to guide light and

enhance the biosensing performance as well as enhanced spectroscopy applications

[104]. Other groups considered the film to play the role of a mirror, and studied

the possibility to have gap modes in the separation between the particles and the

film [105]. The hot spots with high electromagnetic in the gap mode may induce

significant enhancement in the Raman signal, leading to a SERS signal increase of

an order of magnitude, compared to the commonly used aggregates [106]. Finally,

coupled NP/film systems are also interesting for refractive index sensing where a

figure of merit (F.O.M) as high as 16 were recorded in the near infrared (850-1200

nm) for Au rings/Au film systems [107].

1.5 Conclusion

In this chapter we introduced the basic theory behind plasmonics and the for-

mulas explaining the underlying physics. We also discussed the different plasmonic

modes, both delocalized and localized, as well as more complicated plasmonic modes

such as gap modes, Fano resonances, and black modes associated with advanced fab-

rication techniques. The second part of this chapter is at the heart of this thesis,

45


and serves as an introduction to our research topic. The principal studies done by

different groups on coupled NP/film systems are discussed, with an emphasis on

the tunability and the appealing optical properties. Even with all the work already

done on coupled NP/film systems a full understanding of their optical properties

is still not achieved. This motivated us to perform a full systematic study, and to

investigate the possible applications of these interfaces.

46

2 Methods and techniques

2.1 Introduction

To go smoothly through the experimental work done in this thesis, all the

methods and techniques used will be introduced and explained in this separate

chapter. Details of the fabrication techniques used to prepare samples with metallic

and dielectric thin films as well as metallic NPs are discussed here. A presentation of

all the techniques used for morphological characterization of the designed interfaces

such as Scanning Electron Microscopy (SEM) and ellipsometry is also provided, as

well as a complete description of all the methods used for optical characterization

from extinction spectroscopy to Raman spectroscopy.

2.2 Fabrication techniques

2.2.1 Thin film deposition

One of the main objectives of this PhD work is to study the coupling between

metallic film and metallic NPs in a metasurface interface both with and without a

spacer layer. For this purpose, we need to fabricate homogeneous 50 nm Au film,

as well as homogeneous dielectric layers with various thicknesses going from 3 up to

100 nm. Physical vapor deposition is one of the techniques which can be used for

such a controlled thin film deposition.

2.2.1.1 Physical vapor deposition

Physical vapor deposition (PVD) is the process of depositing materials directly

from the vapor phase, it is used to evaporate thin films from few nanometers up to

several micrometers [108] with good adhesion and high purity and homogeneity. One

of its main advantages lies in the wide range of materials which can be evaporated

from metals to dielectrics. PVD is based on heating a source material either with e-

beam heating where the electrons are extracted from a cathode filament and focused

using a magnetic field on a crucible containing the evaporant, or by using electrical

current also known as a Joule heating where the current passes through a metal

crucible and heats the evaporant. Once sufficient thermal energy to bring the source

47

Chapter 2 Methods and techniques

to the boiling temperature is provided, atoms start to evaporate in a uniform and

isotropic manner. It is very crucial that the evaporation process occurs in a vacuum

chamber to avoid all collisions between the evaporated atoms and the atmospheric

gases as the atoms try to reach the substrate. Several parameters control the quality

of PVD such as the pressure, the distance separating the substrate and the source,

and the source temperature.

2.2.1.2 PlASSYS MEB 4000 evaporator

In order to prepare Au films of 50 nm thickness, 3 nm Cr films used for NP

adhesion, as well as SiO2 films with thicknesses varying between 3 and 100 nm,

the PLASSYS MEB 4000 evaporator present in the clean room of our Lab was

used. Metals were heated using the Joule heating via a tungsten crucible, while

dielectrics were heated via e-beam heating. The PLASSYS MEB 4000 is equipped

with two pumps: a primary vane pump, and a secondary turbo molecular pump in

order to increase the pressure inside the system up to 10−7 torr. For monitoring the

evaporation thickness, a quartz crystal oscillator is integrated in the evaporation

chamber. The quartz allows sensitive monitoring of the thickness by measuring the

change in oscillation frequency as the deposition on the crystal face is increased.

2.2.2 Fabrication of Au NPs Via Electron Beam Lithography(EBL)

As the aim of this study is to understand the role and the physics of the coupling

between metal NPs and metal films, it was mandatory for us to prepare identical NPs

on different substrates and interfaces. For this reason, NP deposition via chemical

routes was not very appealing since the spatial distribution of the NPs remains a

challenge. Instead we used electron beam lithography to fabricate reproducible NPs

with exactly specified size and spacing. EBL is a technique which allows imprinting

nanostructures with high accuracy and precision for fabricating patterns going down

to the nanoscale [109]. The main principle behind consists in using the focused

electron beam in a scanning electron microscope across a substrate covered with

an electron sensitive resist layer. The electron beam scans designed patterns with

specific doses and exposure times to ensure the outcome is the desired pattern. The

main advantages of EBL is the high resolution and homogeneous reproducibility

of the nanostructures as well as the high complexity of NPs shape and the wide

range of possible patterns. However it remains a high cost technique which is not

suitable for mass production. This nanofabrication technique has been developed

in our laboratory for more than fifteen years and is well mastered and controlled at

this stage. Fig. 2.1 shows a schematic of the principal steps of the EBL process.

48

2.2 Fabrication techniques

Figure 2.1: Schematic of the elctron beam lithography. 1) Spin coating the glass

substrate with a thin layer electron sensitive resist. 2) Spin coating the sample

with a conductive polymer to avoid the charging effect. 3) Imprinting the designed

pattern with the electron beam. 4) Rinsing with water to remove the conductive

polymer. 5) Development with MIBK/IPA. 6) Evaporation of a 3 nm adhesive

Chromium layer. 7) Evaporation of metal. 8) Lift off using acetone as a solvant

to remove the remaining resist and the metal on top of it, resulting in metal

nanostructures on the desired substrate.

49


Positive e-beam resist deposition

On a clean glass substrate we deposit by spin coating a polymethyl methyl acrylate

(PMMA) layer which is considered as a positive e-beam resist. A positive resists is a

polymer that is weakened and broken into smaller parts with higher solubility when

exposed to an electron beam. In our process PMMA which is a long chain polymer

with 950 k molecular weight is used as a positive resist. The PMMA is dissolved

in the methyl isobutyl ketone (MIBK) solvent, at a concentration of 30 g/L. The

PMMA is then spin coated with a speed of 4000 round.min−1, acceleration of 3000

round.min−1.s−1, and a spinning time of 30 s. These spin coating parameters result

in a PMMA thickness of 150 nm, which is ideal to fabricate NPs with 50 nm height

as experience in our laboratory have shown that the optimized ratio of NPs height

to PMMA thickness is 1:3.

Conductive layer deposition

When performing EBL on a non-conductive substrate such as glass, it is essential to

add a conductive layer in order to avoid accumulation of charge or what is known

as charging effect. This is done by spin coating a thin layer of a conductive poly-

mer “espacer 300Z” at a speed of 2000 round.min−1 and an acceleration of 3000

round.min−1.s−1 for 30 s. The advantage of this polymer is the ease with which it

can be deposited and removed, since it is soluble in water.

Electron beam exposure

The electron irradiation (seeFig. 2.1 part 3) is done using the SEM Hitachi S-3500N

coupled to a beam control system known as the nanometer pattern generation system

(NGPS). The procedure involves several steps. We start by designing the desired

patterns on a 2D graphic software, then we designate the different parameters re-

quired for the exposure such as the dose of irradiation, the exposure time for each

pattern, and the accelerating voltage. It is important to note that several tests are

required before determining the optimized parameters for each new pattern. The

next step is to verify the beam alignment and the correction of the astigmatism.

After which, we move the beam to a Faraday cage used for automated beam current

measuring and compensation and we measure using a Picometer the current inten-

sity at the sample interface. The current is regulated till it reaches 10 pA which is

the current assumed by the software when calculating the exposure time. Finally

we set the magnification to 1000. Finally we launch the lithography and the current

beam starts drawing with the help of an automated XY stage and a beam blanker

the desired design.

50

2.3 Structural characterization

Development of the sample with MIBK/IPA

First we start by removing the conductive layer (in case it was used) by rinsing the

sample with distilled water. After exposing the sample to the electron beam, the

chemical composition of the resist in the exposed areas is modified: the polymer

chains break down and become soluble in an MIBK solution. The development

is then achieved by immersing the sample in a solution of MIBK and isopropanol

(IPA) with 1 to 3 proportion for 1 minute, and rinsing in IPA for another 30 s.

This dissolves the irradiated parts of the resist leaving the PMMA layer with holes

corresponding to the desired lithographic pattern as shown in Fig. 2.1(part 5).

Metal Evaporation

After the holes are patterned on the PMMA, a 3 nm Cr adhesion layer is deposited

using the PLASSYS MEB 4000 before evaporating 50 nm of Au or Ag film as shown

in Fig. 2.1(part 7).

Lift-off

The last step consists in removing the remaining PMMA and lifting off with it the

metal residing on the PMMA as shown in Fig. 2.1 (Part 7). To do so, we immerse

the sample in an acetone solution for at least 3 hours. This is enough to dissolve

and remove the PMMA which was not exposed to the current beam. This final step

leads to Au nanostructures patterns on our substrate.


In optics and nanotechnology, it is critical to monitor and control the morpho-

logical properties of the designed interfaces. Thus structural characterization should

be regularly performed. In this study, it was requested to ensure that the size and

shape of NPs were good, the thickness of the metal and dielectric thin films is well

controlled, and the quality and chemical state of graphene throughout the experi-

ments were not modified. For this reason, we used a Scanning Electron Microscopy

(SEM) to produce high quality images of the NPs, an ellipsometer to measure the

thickness of the thin films, and an X-Ray Photo Electron Spectroscopy (XPS) to

give an insight on the chemical state of graphene layers.

2.3.1 Scanning Electron Microscopy (SEM)

The human eye is able to distinguish two points 0.2 mm apart. Therefore, the

need for observing smaller objects with high resolution led to the development of the

51


optical microscope where a lens or an assembly of lenses could magnify this distance

and allowed the eye to resolve even closer points. Optical microscopes can reach a

magnification of 1000x. They normally operate under visible light with an average

wavelength of 550 nm. The theoretical limit of the resolution is 200-250 nm provided

by the theoretical limit of diffraction d = λ/2N, with λ being the wavelength and

N the numerical aperture. For medical and physical science communities, there is

a need to go beyond this resolution limit. This was made possible with the devel-

opment of the scanning electron microscope (SEM) in the 1950’s. The SEM is a

microscope that uses electrons instead of light. The wavelength of electrons is much

smaller than that of photons, it is around 2.5 pm for 200 keV. Theoretically such a

low wavelength leads to an almost infinite resolution limit, however it is limited by

the objective lens system used in electron microscope. The interaction of the elec-

trons with a sample may provide topographical, morphological and compositional

information from the micrometer going down to the nm scale [110]. It also allows

capturing high resolution 2D images. Since its early development, SEM has been

considered as an essential and un-replaceable technique in morphological characteri-

zation. In this thesis work, SEM was used to produce high quality images to confirm

the size and quality of designed NPs.

The main components of a SEM are illustrated in Fig. 2.2. Electrical current

passing through a tungsten filament forms an electric gun that emits electrons. The

electrons are accelerated by applying a potential difference (1-30 kV), and then

pass through several electromagnetic lenses which control the flow and generate an

electron beam. The beam is focused on a definite spot on the sample’s surface,

the impinging electrons with a specific kinetic energy interacts with the sample and

yields a SEM signal. An incident electron beam drives many forms of emission from

the illuminated surface mainly secondary electrons (SE) and backscattered electrons

(BSE). This requires using two separate detectors and thus two kinds of imaging

can be obtained.

Secondary electrons (SE) imaging When high kinetic energy electrons impinge

on a surface, they extract valence electrons from atoms near the surface of a sample

material. When the energy is sufficient, the electrons are excited and escape the

sample surface. The SE detector uses a positive bias on its front surface in order to

attract the low energy secondary electrons. Secondary electron imaging is surface

sensitive and provides a high resolution. The SE mode is more commonly used,

especially when topographical information is required. Secondary electrons can be

easily deflected by regions where electric charge builds. In non-conducting samples,

this may produce severe image distortion. For this reason previous metallization of

non-conductive samples is usually done to obtain good quality images.

Backscattered electrons (BSE) imaging Backscattered electrons as implied

by their names are incident electrons bounced back from the samples nucleus after

52


Figure 2.2: Illustration of a scanning electron microscope (SEM) showing the dif-

ferent component. Image courtesy of Iowa state University.

elastic collisions. These electrons are scattered at high angles, for this reason the

BSE detector is usually placed directly above the sample as shown in Fig. 2.2. The

quantity of the backscattred electrons and thus the intensity of the BSE signal

depends on the atomic number of the atoms in the studied material. Therefore BSE

images display atomic number contrast where areas of higher average atomic number

are brighter. BSE contains less topographical information however it provides more

information on the chemical composition of different parts of the sample.

Throughout this PhD work, two SEM microscopes were used to characterize the

samples and produce high quality images. The Hitachi SU 8030 and the Hitachi

S3500 N.

2.3.2 Ellipsometry

Ellipsometry is based on the measurement of wave polarization. It is commonly

used in optics with light waves. For reflection ellipsometry, it is not the polarization

of light wave itself but the change in polarization which is of interest. The change

in polarization is represented as an amplitude ratio ψ and a phase difference ∆.

This makes ellipsometry suitable for measuring material properties associated with

a change in the optical response, especially thicknesses of thin films [111]. A typical

53


ellipsometer thus consists of a light source which emits circular or unpolarized light.

Our optical setup consists of:

1) a lamp with a wavelength range of 300 to 800 nm,

2) a linear polarizer to convert the incoming light to polarized light,

3) a compensator or what is also known as a linear retarder that alternates the

polarization state of the wave by retarding the two perpendicular components of the

electric field by different amounts,

4) a sample holder, where the material to be studied is mounted, on this sample

surface a fraction of the light wave is transmitted and another is reflected according

to Fresnel reflection and transmission coefficient,

5) an analyzer which is also a linear polarizer and a detector that measures the

intensity of light arriving from the analyzer,

Figure 2.3: Schematic of an ellipsometer showing the light source, the polarizer,

the compensator, the sample holder, the analyzer and the detector which measures

the intensity transmitted from the analyzer.

Fig. 2.3 shows a schematic of the basic components of an ellipsometer. After per-

forming ellipsometric measurements and knowing the parameters of the polarizer,

analyzer and compensator the amplitude ratio ψ and phase difference ∆ can be

calculated using several methods. Using the ellipsometric parameters ψ and ∆ op-

tical parameters such as the dielectric constant and the thickness of the films can

be calculated. Fig. 2.4 shows the required steps to do the measurements and then

model and fit the results to calculate the dielectric constants as well as the thickness

of the SiO2 thin films.

54


Figure 2.4: Schematic of the steps required to perform ellipsometry measurements.

2.3.3 X-Ray photo electron microscopy (XPS)

A large part of this thesis work concerns graphene and how it can mediate and

enhance optical sensors. It is important to control the quality of the graphene sheets

used, especially when transferred from nickel foil. One of the important parameters

to check is the chemical state of graphene both before and after being transferred on

desired interfaces. To do so, we performed X-Ray photoelectron spectroscopy (XPS)

measurements since it is widely used for monitoring both the oxidation state and the

chemical composition of solids and thin films. XPS is a surface sensitive technique

whose basic principle is to expose a specific sample with X-Ray photons. When the

photons possess energy higher or equal to the binding energy of the electrons in the

atoms they can knock them off on the outer surface of the sample thus emitting

photoelectrons as indicated in Eq 2.1:

EB = hν − Ek + ϕ (2.1)

where EB is the binding energy, Ek the kinetic energy of the emitted photoelectron,

and ϕ the work function whose value is a specific constant which depends on the

spectrometer. The value of the kinetic energy of the emitted photoelectrons is

very specific to the different elements constituting the sample and can be measured

using an electromagnetic detector as shown in Fig. 2.5. Once the kinetic energies

are measured, the binding energy is calculated and the number of electrons with

specific energy emitted per time unit or what is known as the XPS spectrum can be

deduced. The intensity of X-Rays depend on the inelastic mean free path (IMPF)

[112] which is the average distance measured along the trajectories that an electron

beam can travel before its intensity decays to ½ of its initial value . The intensity of

the incident photons decays fast and only shallow atoms or atoms with a distance <

55


Figure 2.5: Schematic showing the main components of an X-Ray phto electron

microscopy.

10 nm from the surface are collected by the detector [113]. Finally it is important to

note that XPS measurements should be performed under high vacuum (10−8 torr) to

avoid the interaction of the negatively charged photoelectrons with the surrounding

atoms in the environment.

2.4 Optical characterization

2.4.1 Surface Plasmon Resonance (SPR) spectroscopy

SPR spectroscopy is a fully established and commercialized technique which

is important for biosensing technology in the areas of biology, biochemistry and

medical sciences because of its real-time, label free and non-invasive nature. SPR

spectroscopy is achieved when incident light is reflected on an interface with a thick

metallic film. In order to overcome the phase mismatch and excite delocalized SPP

modes, light passes through a glass prism at a specific angle of incidence in total

internal reflection. SPP excitation results in light absorption or the attenuation

of reflected light, which yields the SPR spectrum [114]. This phenomenon is very

sensitive to the changes in the medium’s index of refraction which makes it a very

suitable technique for sensing applications especially when molecular adsorption to

the metal interfaces alters the index of the medium. Several measurement modes

can be used depending on the property of interest.

Angle scan mode: in which the system sweeps the angle of incidence at a fixed

wavelength (usually chosen close to the SPP resonant wavelength) and monitors the

reflected intensity.

56


Wavelength scan mode: which measures the reflected intensity for a continuous

wavelength incident laser at a fixed incident angle.

Kinetic mode: in which both the incident angle and wavelength are fixed and the

reflected light intensity is measured as a function of time. The advantage of this

mode is the possibility to monitor in real time the changes in the index of refraction

due to molecule absorption and desorption.

We performed our SPR measurements in a wavelength scan mode using the Kretschmann-

Raether coupling geometry with an attenuated total reflection configuration as

shown in Fig. 2.6. A Fianium supercontinuum laser with a wavelength of 450-1800

nm and a pulse duration of 400 fs was used with an Andor electron multiplying

CCD detector in order to respectively illuminate and collect the reflected intensity.

Fig. 2.6 shows a schematic of the basic parts of the SPR spectrometer which was

used.

Figure 2.6: Schematic of SPR spectrometer in wavelength scan mode.

2.4.2 UV-VIS-NIR spectroscopy

Ultra violet, visible, and near infrared (UV-VIS-NIR) spectroscopy is a tool

used to characterize different optical properties mainly absorption, transmission,

and reflectivity of various materials, coatings, thin films and solutions [115]. It is

based on the interaction of electromagnetic radiation with matter. Molecules in

matter absorb light in form of photons with specific quanta of energy. This leads to

transitions in the energy levels from the ground state to higher excited states. The

absorbed energy can thus be expressed as:

E = hν = hc/λ = hν (2.2)

57


where h is the Planck’s constant, c the velocity of light, λ the wavelength, and ν the

wavenumber. The main features of the absorption band are its wavelength as well

as its intensity. Both are dependent on the energy difference between the ground

electronic state and the excited state. The energy difference is very specific and

dependent on the nature and size of the studied material, which makes it a suitable

technique for identifying and characterizing different plasmonic systems. According

to the Beer-Lambert law, the absorbance depends on the path length or the pene-

tration depth of light into a specific material and the material’s concentration. This

can be understood through the following formula:

A = logI

Io

= εlc (2.3)

where A is the absorbance, ε the absorbance constant, l the path length, c the con-

centration of the studied material, Io the incident beam intensity and I the trans-

mitted beam intensity. UV-VIS-NIR has several measurement modes, the one of

interest for this study is the extinction spectroscopy. As introduced in the previ-

ous chapter, extinction is the sum of absorbed and scattered light which represent

the optical loss. To do so, the sample and the detector should be well separated

and the NA should approach zero. During this research work, we used two UV-

VIS-NIR spectrometer, a home made set-up which was used for all polarization and

angle resolved measurements, and a Nikon microscope which was used for all sensing

measurements since its horizontal sample holder is convenient for adding different

solutions on the samples.

Home made setup

Our home built transmission optical microscope consists of a white light source

focused using a multimode optical fiber and two inverse objective lenses (20x, 0.28

NA) on the surface of a sample mounted on a vertical x-y-z adjustable sample holder.

Collected light is detected by the ocean optics QE65000 detector using optical fiber

and a (20x, 0.42 NA) objective lens. The spectrometer is coupled to a high definition

camera in order to identify different lithographic zones. Based on the optical fibers

and the confocal setup the spatial resolution of the illumination spot is 10µm. The

SpectraSuite software was used to process the collected data and yield the absorption

spectra. In order to measure proper normalized spectra, the SpectraSuite calculates

the absorbance (Aλ) using the following equation:

Aλ = −log10

(Sλ −Dλ

Rλ −Dλ

)(2.4)

where Sλ is the sample intensity at a wavelength λ, Dλ is the dark intensity at a

wavelength λ, and Rλ is the reference intensity at a wavelength λ. The dark signal

is measured when the white light source is turned off, the reference is measured with

the light on regions of the sample without NPs, while the sample measurement is

done on the NPs. This ensures that the measured signal is only that of the NPs

58


and that it is properly normalized. When using this formula for NPs in a solution

the resultant spectrum corresponds to the extinction. However for the case of NPs

on a substrate, one should be careful since the measured spectra does not perfectly

correspond to the extinction, as reflection caused by the substrate also contributes.

The real measured quantity is -log(T/To) where T is the transmitted light on regions

with NPs and To the transmitted light by the substrate but we will refer to this as

extinction spectra throughout the manuscript.

In addition to the basic components of the spectrometer, a combination of a

polarizer and an analyzer added to the setup allows measurements with different

polarization conditions. The optical setup which was used was modified to tilt

the sample, allowing us to investigate illumination and collection angles beyond the

traditional normal incidence configuration, up to 60° under TM and TE polarization.

All the different components of the setup are schematically shown in Fig. 2.7.

Figure 2.7: Scematic of the home made extinction spectrometer [12].

59


Nikon Eclipse TE2000-U

Figure 2.8: Scematic of the TE2000-U microscope, showing the different com-

ponents of the microscope. The T-DH dia illuminator 100W, LHS-H100P-1

12V100W halogen lamp, TE2-PS100W power supply, a rectangular stage used

as a sample holder, a binocular eyepiece tube, a system condenser, a 0.45 objec-

tive, cords ...

Fig. 2.8 shows the different parts of the Nikon microscope which allows performing

extinction measurements in transmission. Similar to the home made setup the basic

parts of the spectrometer are the illuminator, a condenser, an x-y controlled stage

used as a sample holder, a microscope, a power supply, and an eye piece tube which

can be replaced by a high definition camera. We used an objective lens with a NA

of 0.45 and a 20 x magnification throughout all measurements. The same ocean

optics QE65000 detector and SpectraSuite software used in the home made setup

were also used on the Nikon setup.

2.4.3 Raman Spectroscopy

Raman scattering is based on molecular deformations in the electric field Edetermined by molecular polarizability ❛. The laser beam can be considered as an

oscillating electromagnetic wave. Upon interaction with the sample an electric dipole

moment−→P = ❛

−→E is induced and the molecules are considered as oscillating dipoles

vibrating with characteristic frequency νm. A molecule with no Raman-active modes

absorbs a photon with the frequency ν0. The excited molecule returns back to the

same basic vibrational state and emits light with the same frequency ν0 as the

excitation source. This type of elastic interaction is called Rayleigh scattering and

60


it constitutes 99.99% of the total interactions [116]. The other type of interaction

is the inelastic Raman scattering which can be of two types, a Stoke or anti-Stoke

one. When a photon with a frequency ν0 is absorbed by a Raman-active molecule,

with the molecule being at the basic vibrational state athe the time of interactiont:

part of the photon’s energy is transferred to the Raman-active mode with frequency

νm, while the resulting frequency of scattered light is reduced to ν0 − νm. This

Raman frequency is called the Stokes frequency. When a photon with frequency ν0

is absorbed by a Raman-active molecule, which at the time of interaction is already

in the excited vibrational state excessive energy of the excited Raman active mode

is released, the molecule returns to the basic vibrational state and the resulting

frequency of scattered light goes up to ν0 + νm. This Raman frequency is called

the anti Stokes frequency. This process is called the Raman effect, first observed

by Chandrasekhara Venkata Raman in 1928 [117] .The different interaction schemes

are illustrated in Fig. 2.9.

h

Figure 2.9: Quantum Energy Transitions for Rayleigh and Raman Scattering.

Raman scattering occurs rarely and only 0.001% of the interactions are inelastic.

For this reason measuring the Raman scattering can be challenging in many cases.

Moreover Raman scattering is very sharp compared to fluorescence, it allows finger-

printing of single molecule. The Raman microscope which we used is the Jobin-Yvon

Horiba (LABRAM) consisting of sevral parts allowing detection and filtering of the

Raman signal:

1) a helium Neon excitation laser with a 632.8 nm wavelength is coupled to the

microscope,

61


2) a band-pass filter is used to remove undesired wavelengths, and a collimated beam

is reflected off the Raman filter towards the objective,

3) a beam splitter which can be placed between the filter and the objective which

directs a part of the light towards a high definition camera in order to allow exact

detection of desired zones,

4) an objective lense used to focus the laser onto the sample’s surface and to collect

the backscattered Raman signal. Since the Rayleigh scattering constitutes 99.99%

of the overall signal, a Raman filter is necessary to remove this undesired signal and

to allow only Raman signal,

5) a holographic grating (with two options: a 600 lines/mm and a 1800 lines/mm)

disperses it and it is then imaged by a Peltier cooled CCD detector (1024x256 pixels)

of 16 bits dynamic range,

6) an Olympus optical microscope with 3 different objectives (10x NA 0.25, 50x NA

0.8, 100X NA 0.9),

7) an 800 mm path monochromator,

Furthermore the LABRAM system can be modified to allow absorbance measure-

ments in reflection. In order to perform such measurements the excitation laser is

replaced by a white light source and the beam splitter position is changed to di-

rect the light towards an ocean optics QE65000 detector. This configuration of the

LABRAM was used to perform reflection measurements since the previous UV-VIS-

NIR spectrometers described earlier are only adapted to work in transmission.

2.5 Conclusion

The main purpose of this chapter is to explain the different methods and tech-

niques we used to perform our experimental results. A brief introduction to the

theory behind each technique is presented, followed by an explanation of the details

and parameters of the different setups we used.

62

3 Plasmonic mode interferences andFano resonances in MlMnanostructured interface

3.1 Introduction

Investigation of coupled NP/film systems was led by several groups either with

the NPs fabricated directly on the metallic film or in the presence of a spacer layer.

Dielectric materials were commonly used as spacer layers, different materials and

thicknesses have been investigated. Some groups performed theoretical studies where

the spacing was simply air [118, 84], others investigated both experimentally and

theoretically different dielectrics with various thicknesses: 20 nm SiO2 [32, 34], 50

nm ITO[33]. Even with all the work already done, an engineered control of the opti-

cal properties of such systems is very challenging since the exact underlying physics

and mechanisms are still not fully understood. A complete comprehensive numerical

and experimental study investigating the effect of the NPs diameters, periodicities,

and more importantly the thickness of the spacer layer is still missing. For this

reason, in the following two chapters, we will present a part of this PhD work which

aims to fill this gap through a full systematic treatment of the problem. Samples

made of Au NPs deposited on a Au film with SiO2layer have been fabricated and

characterized. To allow a global understanding of these systems , NPs with five dif-

ferent diameters and three different periodicities were fabricated via EBL onto Au

film coated with SiO2. We start of by comparing the effect of introducing a metal

film on the optical and plasmonic properties when compared to similar systems with

the same NPs fabricated directly on glass. After establishing a good understand-

ing of the influence of the metallic film, we increased the complexity of the studied

system by adding an ultra thin SiO2 spacer layer sandwiched between the NPs and

the film in order to establish our MIM structure. For this new system, the optical

properties were massively altered which required a full numerical and experimen-

tal investigation to understand the new features, their optimization and potential

applications. Numerical simulations were performed by our colleague Dr Gaëtan

Lévêque from the Institut d’Electronique de Microélectronique et de Nanotechnolo-

gie. Studying MIM structures is very important, especially with the possibility to

control the transmission, absorption, and reflection of these systems making them

appealing metasurfaces [35] for different optical applications.

63

Chapter 3

Plasmonic mode interferences and Fano resonances in MlM nanostructured

interface

3.2 Sample fabrication and structural characterization

Studying hybrid metal-insulator-metal (MIM) systems is rather difficult due to

the complexity of the underlying physics. Treherefo, we decided to build our work

and our understanding of these systems step by step.We first compared the struc-

tures with NPs directly on a metal film to the one with the NPs on glass. Then we

compared the previous two systems with a more complicated MIM structure where

the insulator is a SiO2 dielectric spacer layer. To do so, we fabricated using EBL

identical Au NPs of different sizes on the three different interfaces. We performed

SEM images on all the samples to ensure that the NPs exhibit exactly the desired

size and spacing, as well as ellipsometry measurements to confirm the thickness of

the SiO2 layer.

3.2.1 Sample Fabrication

In order to prepare the different interfaces, we cleaned four glass substrates by

putting them in a solution of distilled water and Decon Neutracon detergent in an

ultrasonic bath at 50°C for 5 minutes. Later we rinsed the samples with distilled

water and immersed them in isopropanol solution and again in the ultrasonic bath

at 50° for 5 minutes. Using physical vapor deposition (PVD), we deposited on three

of the clean glass substrates a 50 nm Au film and on two of the Au coated samples we

deposited using PVD a thin SiO2 layer of 3 and 6 nm thickness respectively. After

preparing the different substrates we deposited on them using EBL arrays of Au

NPs with 50 nm height and 3 nm Chromium adhesion layer to ensure the NPs are

well attached to the substrate. Fig. 3.1 shows a schematic of the different designed

interfaces.

Ordered square arrays of NPs with a 50 x 50 µm dimension were prepared

using the EBL technique. The different arrays represent NPs of different sizes and

spacings. Five different diameters 200, 170, 140, 110, 80 nm and 3 different center-to-

center distance or periodicites 300, 450, and 600 nm were designed. Fig. 3.2 presents

a dark field microscopy image showing twelve of the fifteen arrays with artificially

added axis indicating the diameter on the x-axis and the periodicity on the y-axis.

It is important to note that this lithography design was used throughout the whole

thesis.

3.2.2 Structural Characterization

Since the optical properties and the plasmonic modes can be highly dependent

even on the smallest variations in the diameters and periodicities of the NPs, it

is crucial to ensure that the sizes of the NPs fabricated on the different samples

are identical. To do so, we used the same recipe while doing EBL. This included

several parameters, most importantly the thickness of the PMMA, the design of the

64

3.2 Sample fabrication and structural characterization

Figure 3.1: Schematic of the designed interfaces with Au NPs on top of a) glass

substrate, b) 50 nm Au film on a glass substrate and c) SiO2 layer/Au film on a

glass substrate.

NPs and the dose and exposure time of the electron beam. However even when

controlling all these parameters, it is necessary to confirm via SEM the quality of

the lithography, to compare the sizes of the NPs and to ensure that the samples

are not contaminated and in good condition. This was done on all the samples

prepared and only high quality samples were later used for optical measurements.

In Fig. 3.3, a collection of SEM images is presented: a) showing a full array of NPs,

b-d) images with higher magnification showing the different periodicities e-i) images

of parts of the arrays with 300 nm periodicity and different NP diameters, j-n)

single NP images with different diameter measurements. An overall observation of

the collected SEM images demonstrates that the periodicities of the fabricated NPs

are in a good agreement with the expected ones with a maximum uncertainty of ±2 nm. Similarly for small NPs with diameters ≤ 140 nm the uncertainty on the size

does not exceed ± 2 nm.

However for the bigger diameters with 170 and 200 nm, the uncertainty was

slightly higher and reached ±4 nm. The second parameter that should be controlled

and monitored while fabricating the samples is the thickness of the deposited films.

Using PVD, the uncertainty level could go up to ± 3 nm. The Au film is 50 nm which

is considered thick enough so that a 3 nm discrepancy has no or very minor effect

on the optical properties. On the other hand, the SiO2 films studied in this chapter

are ultra thin with 3 and 6 nm thicknesses. For such ultra thin dielectric spacer

layers, even the slightest discrepancy between the expected and the experimental

thickness value can drastically change the optical properties as we will discuss later.

For this reason, ellipsometry measurements are necessary. It was very challenging

to prepare dielectric spacer layer with the desired thickness and with a high degree

of homogeneity especially for the 3 nm spacer. In the process several samples with

65

Chapter 3


interface

Figure 3.2: Dark field microscopy image showing twelve Au arrays, the x-axis dis-

plays the diameters of the NPs in the arrays: 200, 170, 140, 110 nm, and the

y-axis displays the three different periodicities 300, 450 and 600 nm.

thicknesses between 1.5 and 3 nm were prepared before achieving an almost homo-

geneous 3 nm spacer. The thickness and homogeneity for the 3 nm was confirmed

by performing ellipsometry measurements on several random spots on the sample.

Similarly for the 6 nm sample ellipsometry measurements confirmed the thickness

and the homogeneity, the thickness was also later confirmed with a cross compari-

son with numerical results. Even after performing ellipsometry measurements, the

results for ultra thin nm spacers remain doubtful since the uncertainty of the ellip-

someter is ± 1 nm which is relatively high when the desired thickness goes down

to 3 nm. It was important to discuss these unsuccessful results, to understand the

limit of the dielectric thickness we could achieve.

3.3 Influence of the metallic film on the plasmonic

properties of metal NPs

Going through recent research on plasmonics numerous studies on coupling

between NPs can be found[119], mainly investigating the changes in the plasmonic

properties when simple and complicated patterns of NPs are brought close together[120].

Research on nano-antennas, their application and how they can be used in quan-

tum optics is also very popular. However less attention was given to the influence

induced by a metallic film placed in the vicinity of metallic NPs. For this reason, in

the following section, we compare the plasmonic modes of two systems: one with Au

NPs placed directly on glass and the second with the NPs patterned on a 50 nm Au

film deposited on a glass substrate. The presence of the film drastically changes the

optical properties so that new modes are observed and previously existing modes

66

3.3 Influence of the metallic film on the plasmonic properties of metal NPs

Figure 3.3: SEM images of a) full array, b-d) higher magnification showing the

different periodicities e-i) parts of the arrays with 300 nm periodicity and different

NP diameters, j-n) single NP with different diameter

are hybridized and shifted. From an applied point of view, the new system with the

Au film shows a high potential for applications in the low wavelength range as it

will be presented in the following sections.

Extinction measurements Suspended NPs in a solution exhibit a longitudinal

dipolar mode with the charges symmetrically separated as discussed in the previous

sections. When the same NPs are placed on a dielectric substrate, the plasmonic

mode is shifted to the red and a slight asymmetry is induced in the charge distribu-

tion due to the interaction with the substrate [118]. The longitudinal dipolar mode

for NPs on a dielectric substrate is characterized by two hot spots on the low corners

of the NPs, i.e on the substrate surface [118]. However when the NPs are placed on

a metallic film, the dipole-film interaction[84] and the hybridization caused by the

presence of the film, as well as the coupling between localized and delocalized plas-

67

Chapter 3


interface

monic modes might significantly change the optical properties of the system[45, 44].

To investigate these modified properties, we measured the extinction spectra of the

two systems with and without a gold film.

Figure 3.4: Extinction spectra of NPs with five different diameters: 80, 110, 140,

170, and 200 nm deposited on a) glass substrate b) 50 nm Au coated glass sub-

strate.

Fig. 3.4 shows the extinction spectroscopy for NPs with a constant periodicity of

300 nm and 5 different diameters on a glass substrate, and a 50 nm Au film/glass

substrate (see Fig. 7.4 a and b). For the particles on glass, a broad mode is observed

in a wavelength window of 600 to 750 nm. The response of this mode to the increase

in diameter is in good agreement with that of a typical LSP mode where an almost

linear red shift is observed. Identification of this mode as a dipolar longitudinal

LSP mode is not very challenging based on both: its red shift versus diameter, and

on the fact that it was commonly reported in the literature for similar NPs on a

glass substrate. However the extinction spectra for the NPs directly deposited on

the Au film case are less common. We indeed observe a mode at a low wavelength

around λ1 = 520 nm and a broad and not well defined band at λ2 = 770 nm. A

remarkable feature of the low wavelength mode is that it is almost independent of the

NPs diameter as clearly seen in Fig. 3.4. SPP modes are independent of diameters

however it is not expected to observe an SPP mode at such a low wavelength and

under normal incidence. In order to understand the nature of this mode compared

to the longitudinal LSP mode on the glass substrate, we plotted the change in the

resonant wavelength versus the diameter for two different periodicities as shown in

Fig. 3.5. In Fig. 3.5 (a) the longitudinal LSP mode for the NPs on glass substrate

shows a clear red shift > 150 nm when the NPs size increases from 80 to 200 nm.

The shift cannot be considered perfectly linear over all sizes of NPs and the results

can be split into two regimes: one for small NPs (80-110 nm) and another one for

large NPs (140 to 200 nm) with each regime exhibiting its own linear slope. This

68

3.3 Influence of the metallic film on the plasmonic properties of metal NPs

dipolar LSP mode is red shifted when the periodicity increases as shown in Fig. 3.5

(a). It is known that a dipolar LSP mode only depends on the diameter of the

NPs. However we discussed in Chapter 2 that for a closely spaced array of NPs,

the ensemble measurement depends both on the NP diameter and the dipole-dipole

interaction between neighboring NPs. Dipole-dipole interaction can be both in-

phase and out-of phase: in phase interaction induces wavelength red shifts in the

ensemble measurements compared to the single NP resonance while out of phase

resonance induces a blue shift.

Eventhough we did not perform single NP measurements, we can still deduce

the kind of interference between neighboring NPs based on the resonance shifts when

the periodicity is changed. When the periodicity increases, from 300 to 450 nm the

strength of the interaction decreases, and a red shift is induced in the plasmonic

resonance. This red shift indicates that the dipole-dipole interaction in our arrays

is out of phase. An out-of-phase coupling would decrease the oscillation of the

electrons and thus requires higher energies for resonance which induces a blue shift

in the resonant wavelength. When the periodicity increases, the coupling effect

decreases since the interaction scales as d−3 with d is the distance between two NPs.

Therefore, a recorded red shift versus an increase in periodicity indicates that as

the interference between NPs decreases, it requires less energy to excite the LSP

mode. This confirms that the interaction between neighboring dipoles in our system

is out of phase. On the other hand, when examining the case of a Au film in Fig. 3.5

Figure 3.5: Resonant wavelength versus the change in the diameter for two differ-

ent periodicities: 300 nm (red) and 450 nm (green) for the a) glass substrate, and

b) Au film.

(b), one can confirm that the 520 nm mode is not a grating induced SPP since it

is almost independent on the periodicity with the LSPR wavelength ∼520 nm for

both the 300 and 450 nm periodicities. The resonance slightly red shifts with only

few nms as the diameter or periodicity increases. A notable remark is that also for

this mode the NPs seem to divide into the same two regimes in a similar qualitative

69

Chapter 3


interface

behaviour compared to that of NPs on glass. In a study on plasmonic modes of gold-

nanoparticle arrays on thin gold films, Hohenau and Krenn [30] observed a similar

mode at 520 nm for Au NPs deposited directly on Au film and they suggested that it

could be a vertically oriented dipole LSP resonance located on the NPs and scatters

to higher energy SPPs. However further investigation is required to define precisely

the origin of this low wavelength mode. The common model to understand metallic

NPs in the vicinity of a metallic film is the dipole-surface interaction model [84],

where the metallic film plays the role of a mirror and the dipole LSP in the NPs

interact with its image dipole induced by the film, this interaction shifts the LSP

mode to the red. The qualitative behaviour of the period independent 520 nm mode

confirms its nature as a dipolar localized mode. However from a physical point of

view it cannot be the same longitudinal mode observed on a glass substrate, since

the latter is expected to red shift due to the dipole-film interaction and is probably

not observed in the Au film system because it is shifted beyond the wavelength limit

of our spectrometer. Also it should not be the vertical dipolar LSP mode since this

mode cannot be excited in normal incidence due to its symmetry. Identification of

the exact nature of this mode cannot be fully achieved using experimental data. For

this reason numerical simulations are required and specifically electric field maps

since it can give an insight on the charge distribution in the NPs.

Electric field maps The electric field map was computed at the resonant

wavelength of 520 nm using the Green’s tensor formalism for a 200 nm Au NP

deposited on a 50 nm Au film. The Au optical constants were taken from John

and Christy[121]. Fig. 3.6 shows the distribution of the electric field inside a vertical

section in the polarization plane of a 200 nm Au NP. Bright colors indicate regions

of high intensity of the electric field and the green arrows represent the real part of

the electric field.

Figure 3.6: Computed distribution of the electric field inside a vertical section of

a 200 nm Au NP on a 50 nm Au substrate at a wavelength λ= 520 nm. The green

vectors show the real part of the electric field.

70

3.4 Ultra thin 3 nm SiO2 spacer layer

The electric field map reveals that this low wavelength mode is a dipolar LSP whose

hot spots are pushed to the top corners of the NPs, probably due to the metallic

film playing the role of a mirror. The contact of the poles of this dipolar mode with

air which is a low index dielectric induces the blue shift towards 520 nm compared

to glass substrates [122]. This low wavelength mode is very interesting for several

reasons:

1) It is one of the few if not the only plasmonic mode which is almost independent

on both the NPs size and the periodicity of the array.

2) It is located at a significantly low wavelength region compared to plasmonic modes

of comparable dimension which gives an accessibility to new applications in this low

wavelength range.

3) Its nature with the hot spot on the top of the NPs i.e on the air interface is

very appealing for refractive index (RI) sensing as well as surface enhanced raman

spectroscopy (SERS).

3.4 Ultra thin 3 nm SiO2 spacer layer

We discussed in the previous section the interesting optical properties associated

with a coupled NP/film system. However the degree of control and tunability of

such a system seems to be very limited. For this reason, we will investigate a more

complicated case with a dielectric spacer layer sandwiched between Au NPs and

Au film since those systems are more tunable and show more resonances compared

to the simple NP on film case. Several theoretical studies have shown that a gap

mode exists between metal NPs and metal film when they are no longer in contact

and a distance d is separating them [118, 84]. Gap modes are very interesting for

sensing, SERS, light harvesting and much more applications since they confine a

high intensity of electric field in a small region. In their study, Lévêque and Martin

[84] showed that for an Au cubic NP with a diameter of 110 nm and a height of

50 nm placed above a 50 nm Au film, the intensity of the electric field in the gap

increases as the separation d decreases. For a distance d around 10 nm the intensity

can go up to 1200 times the intensity of the incident field and when this distance is

further decreased to ultra thin spacing < 1 nm, the intensity can even go up to 5000

times. Encouraged by this theoretical work, we decided to experimentally study

the optical properties of ultra thin spacers. Experimentally the NP cannot simply

elevate above the film as in the simulations. For this reason, we used SiO2 as a spacer

dielectric material. Thin films below 1 nm can only be experimentally achieved

through atomic layer deposition (ALD), a technique which was unfortunately not

available to us. Therefore we had to prepare thicker SiO2 thin films using PVD.

We tried using the PlASSYS MEB 4000 evaporator present in our lab to prepare

a SiO2 layer with a thickness going down to 3 nm on top of the Au film. Several

samples with these parameters were fabricated. Ellipsometry measurements on the

71

Chapter 3


interface

set of prepared samples showed that for such a thin spacer the thickness is highly

inhomogeneous and the film is not continuous. However we decided to proceed and

prepare NPs on top of the substrate using EBL. The extinction spectra for the 300

nm periodicity of the prepared sample are shown in Fig. 3.7 and they are very similar

Figure 3.7: Extinction spectra under normal incidence for Au NPs with 300 nm

periodicity and five different diameters on a 3nm SiO2/ 50 nm Au film.

to the case of NPs deposited directly on a thin film. Unexpectedly there is no sign

of a gap mode. This is probably due to the inhomogeneity and bad quality of the

SiO2 layer. Indeed it is almost impossible to check the exact thickness of the SiO2

spacer in the region under the NPs. This forced us to proceed our investigations

with relatively thicker thicknesses as a compromise between the decreased intensity

of the enhanced electric field and good quality controlled measurements.

3.5 Optical properties of MIM structure with 6 nm

SiO2 spacer layer

Metal-insulator-metal (MIM) systems exhibit a rich underlying physics leading

to a high degree of tunability of their spectral properties. We performed a sys-

tematic study on a metasurface consisting of nanostructured MIM system with a

thin 6 nm dielectric, as this was the minimum thickness we could achieve for a

homogeneous dielectric layer as tested by ellipsometry measurements. We showed

how the nanoparticle sizes and excitation conditions may lead to the tunability and

coupling/decoupling of localized and delocalized plasmonic modes. We also experi-

mentally evidenced a tunable Fano resonance in a broad spectral window 600 to 800

nm resulting from the interference of gap modes. By varying the incident illumi-

nation angle, the resulting shifts in the resonances give the possibility to couple or

decouple the localized and delocalized modes and to induce a strong change of the

asymmetric Fano profile. All these results are confirmed with a crossed comparison

72

3.5 Optical properties of MIM structure with 6 nm SiO2 spacer layer

between experimental and theoretical measurements, confirming the nature of the

different modes. The high degree of control and tunability of this plasmonically rich

system paves the way for designing and engineering of similar systems with numer-

ous applications. Variations in NPs diameters and periodicities as well as the spacer

layer thickness and excitation conditions have a direct influence on the wavelength

of plasmonic modes, as well as their nature. Our nanostructured (MIM) interface

consisted of a 6-nm-thick SiO2 spacer sandwiched in between a grating of Au NPs

designed by EBL using the same pattern introduced previously and a 50-nm-thick

Au film. It is a rich plasmonic playground supporting several plasmonic modes of

different nature in a relatively small spectral window.

3.5.1 Experimental results

In order to study the far field behaviour of our MIM structure, we performed

extinction spectroscopy on the different arrays of NPs. Fig. 3.8 shows extinction

spectra in transmission and under normal incidence for NPs with three different

grating periodicities (300, 450, and 600 nm) and five different diameters (80, 110,

140, 170, 200 nm). From the extinction results, one can conclude that the MIM

structure with a thin insulator layer is a plasmonically rich system supporting several

plasmonic modes of different nature as their lineshapes show, in a relatively small

spectral window.

A low wavelength mode at 520 nm similar to the one described in the previous

section is observed for the three different grating periodicities. The position of the

peak was not altered by the presence of the insulator layer, and its nature was not

changed: the metallic film was still able to play the role of the mirror pushing the hot

spots of the dipolar LSP to the top corners of the NPs resulting in a low wavelength

resonance.

Besides this mode several other resonances whose wavelength depends both

on the periodicity and diameter are present in the system. In order to get an

insight on their nature, we studied their size and grating dependency. The diameter

and periodicity dependency is plotted in Fig. 3.9. The experimental results modes

observed are presented in Fig. 3.9 (a):

1) The 520 mode is again almost independent of period with a slow slope linear

dependency on the diameter.

2) A mode independent of the diameter observed respectively at 560 and 620 nm

wavelengths for the 300 and 600 nm grating periodicities (see Fig. 3.9 b), from the

dependency on the grating periodicity and Eq 1.9 we can conclude that this is a SPP

mode. A question that will remain unanswered till later through this chapter is why

no SPP mode was excited for the 450 nm periodicity and what is the nature of the

excited SPP modes observed for 300 and 600 nm grating. If we compare with the

system without spacer layer, we notice that no SPP mode was excited in the latter.

For NPs directly on film, the only SPP mode which can be excited is the air SPP i.e

73

Chapter 3


interface

Figure 3.8: Extinction spectra in transmission and under normal incidence for dif-

ferent diameters of Au NPs (80, 110, 140, 170 and 200 nm) with center-to-center

distances of a) 300 nm, b) 450 nm and c) 600 nm.

the SPP propagating on the metal/air interface. So far all our measurements were

performed under normal incidence and under these conditions the propagating SPP

mode cannot be excited. When a dielectric layer is added as in the MIM case, a

SPP mode can be excited both at the metal/glass interface and at the air interface

as well as coupled symmetric and antisymmetric SPP modes. In addition to that,

the high index of refraction of glass changes the effective index of refraction of the

interface and the phase matching condition required for exciting an SPP mode is

satisfied even under normal incidence.

3) The third and last mode observed in our system is very interesting as we observe

a highly asymetric mode in a wavelength window of 600 to 800 nm. Typically

conventional plasmonic modes are characterized by a symmetric lorentzian form.

However with the advancement in the fabrication techniques more complicated NPs

and patterns have been designed and asymetric Fano profiles have been increasingly

74


reported in plasmonic systems [123, 124, 70, 75, 76]. As defined in Section 1.3.8, a

Figure 3.9: Resonance wavelength versus the diameter of NPs for the three differ-

ent periods (300, 450 and 600 nm): a) for the localized mode at low wavelengths

around 520 nm, b) for the delocalized mode at higher wavelengths between 560

and 620 nm and c) for the Fano resonance.

Fano resonance is the result of the coupling between a localized narrow mode and a

continuum. It is crucial to understand in our system the origin of the two coupling

modes.It can be seen in Fig. 3.9 (part c) that the Fano resonance wavelength is

linearly dependent on the NPs diameter and blue shifts as the periodicity increases.

4) Besides as expected, a parallel dipolar mode can be seen for the smallest NPs of

diameter 80 nm around a wavelength of 820 nm. This mode is red shifted as the NPs

diameter increases, which explains why it is not observed for NPs > 80nm, since

their resonance is outside the spectrometer accessible wavelength range. However

the interesting feature of this mode is its lineshape, which also corresponds to a

Fano profile. This Fano mode is probably of a different nature of that observed for

the larger NPs since they are characterized by different lineshapes.

Finally it appears highly important to understand the nature of the SPP modes, as

well as the origin of the two coupling modes resulting in a Fano resonance. However

this cannot be achieved experimentally and numerical simulations are necessary for

a complete analysis.

75

Chapter 3


interface

3.5.2 Numerical simulations

For a complete analysis of the measured plasmonic modes and the origin of the

observed Fano resonance, numerical simulations were performed by our colleague

Gaëten Lévêque. We investigate in this part the plasmonic properties of a gold

nanocylinder deposited on a dielectric spacer of varying thickness, over coating a

50-nm thick gold film on a semi-infinite glass substrate. The particle has a fixed

thickness of 50nm, and a range of different diameters. In a first part, the single

particle is investigated. First, full numerical simulations are performed using the

Green’s tensor formalism, and then a simplified analytical model is proposed to

understand the physical origin of the localized plasmon modes. In the second part,

the complete periodic system is investigated. Again, we use the periodic Green’s

tensor formalism to compute exactly the response of the nanostructured interface.

These simulations reveal sharp Fano profiles in the transmission spectra around the

resonance wavelengths of the localized plasmon modes. Interestingly, these modes

appear symmetric in the reflection spectra. We propose then a simplified analytical

model which allows reproducing this behavior for reflection and transmission spectra.

First we have investigated using the Green’s tensor formalism the plasmon properties

of a single gold cylinder separated by a 6-nm-thick silica spacer from a 50-nm-thick

gold film. The cylinder presents a height of 50 nm and a diameter d ranging between

80 nm and 200 nm, and is excited from the air by a plane wave in normal incidence.

The extinction cross-sections are plotted as a function of the wavelength on Fig. 3.10

(a). Several resonances, corresponding to the excitation localized surface plasmon

modes in the coupled particle-film system, can be observed. The nature of these

modes can be elucidated by plotting the distribution of the electric field just under

the bottom surface of the particles or in the polarization plane of the incident wave.

1) The mode (D) at 825 nm for d=80 nm is the dipolar plasmon parallel to the film,

as shown on Fig. 3.10 (b). It is known that this mode is red-shifted when the particle

diameter increases, hence for larger diameter its wavelength will be beyond 900 nm.

2) We see that the mode at 520 nm (labeled 1) is almost diameter-independent.

That plasmon modes has already been observed in several other experiments and

publications, and is characterized by two maxima of the field at the top edges of the

particle, making it looking like a particle dipole pushed at the outer particle surface.

3) The modes with wavelength between 600 nm and 800 nm (labeled 3) for d =

110, 140, 170, 200 nm correspond to gap modes where light is concentrated in the

silica spacer under the cylinder particle. The distribution of field on Fig.1(b) for

the 200nm diameter particle shows a clear interference pattern, originating from

the fact that the area of the spacer under the nanoparticle behaves like a resonant

cavity for the propagative surface plasmon modes. Indeed, the incidence plane wave

excites by diffraction a delocalized metal-insulator-metal SPP (MIM-SPP) under the

particle which, due to the strong impedance mismatch at the nanoparticle edges,

bounces back and forth under the cylinder and form resonant patterns for specific

wavelengths

76


Figure 3.10: (a) Absorption spectra of a single cylinder gold particle of varying

diameter on top of the multilayer substrate. The excitation is a TM-polarized

plane wave in normal incidence, from the air side. (b) Distribution of the electric-

field intensity for modes indicated on spectra (a): the distribution is plotted in the

polarization plane for modes (1) and (d), and in a plane parallel to the substrate

just under the bottom surface of the cylinder for modes (2) and (3). Maps with

the same label on the spectra (a) have similar field distributions; the green arrows

on maps (1) and (d) show the real part of the electric field. (c) Amplitude of the

field scattered at infinity in the direction of the transmitted incident wave (0°) for

particles with diameters 80 nm (black) and 200 nm (green), which slowly vanishes

asymptotically.

Finally, Fig. 3.10 (c) puts into evidence the amplitude of the field scattered in the

silica, normally to the interface. So no interference with the incident plane wave is

taken into account here. An important fact here is that the Fano profile occurring

at 720 nm for the 200 nm particle is already present in the scattered field, which

means that it does not originate in the interference with the continuum of incident

plane wave, but must find its explanation in the inner properties of the particle-film

system. Notice that the Fano profile is not only found in the scattered field coming

from the transmission but as well from the reflection on the substrate. However, for

the 80 nm particle, the profile of the mode around 825nm is completely symmetric,

which means that its origin is very different from the 200nm particle gap mode at

720 nm, and probably comes from the interference with the directly transmitted

plane wave through the substrate.

Even though single NP simulations donnot give a complete description on the

physical problem, but it is very important especially for understanding the nature

of the different localized resonances. However they do not allow to investigate delo-

77

Chapter 3


interface

calized modes propagating on the surface. To do so, first we briefly investigate the

delocalized modes supported by the substrate (dispersion curve and field profile),

and how those modes can be coupled to an incident plane wave by a grating weakly

interacting with it. Indeed, the substrate alone (without the gold nanocylinders)

supports two SPP modes, one for which the light intensity is maximum on the air

side (actually just above the thin overcoating glass layer), and the other where the

light intensity is maximum on the silica side of the gold film (see the field profile on

Fig. 3.11 a). In the weak coupling regime, the excitation condition is obtained by

folding the dispersion curves of the two SPPs in the first Brillouin zone, following:

k2SP P = (kx +m2π/L)2 + (n2π/L)2 (3.1)

where kSP P is the wavector of the considered SPP modes, m and n are integers

and L the periodicity of the grating. The incident planewave is tilted along one

of the grating main axis, labeled x here, and has a parallel component kx of the

wavevector along the interface. In order to compare with the first set of experiments

(seeFig. 3.8), Fig. 3.11 (a) presents the evolution of the SPPs excitation wavelengths

in normal incidence (kx = 0) as a function of the grating period and to the right part

of the figure the profile of the electric field for different SPPs. We can observe that

for the experimental periods of 300 and 450 nm, only the delocalized plasmon SPP

(m=1, n=0) on the silica side (in red) the i.e the maximum profile of the electric

field is on the glass as shown in part 1 of Fig. 3.11 (a) can be excited at respectively

560 and 720 nm. The mode at 560 nm for 300 nm periodicity is in perfect agreement

with the observed delocalized modes in the experimental results. While for the 450

nm periodicity the resonance wavelength of the SPP at 720 nm spectrally overlaps

with the Fano resonance, this explains why it was not observed experimentally.

This overlap of two narrow modes, the silica SPP and the gap mode, resulted in

sharper Fano resonance compared to those measured for the other periodicities. This

highlights the importance of the nature and intensity of the modes interfering in a

Fano profile on the sharpness and degree of asymmetry of the Fano profile. Whereas

the air-side SPP (m=1,n=0) can be excited for a grating period of 600 nm at ❧=625

nm as shown by the solid black line in Fig. 3.11 (a) with the electric field profile

presented in (3), close to the experimental wavelength of 635 nm (see Fig.3.9 b).

Since SPP modes highly depend on the angle of incidence, it could be very in-

teresting to investigate these modes for angles beyond the normal incidence Fig. 3.11

(b) shows the folded SPP curves in the first Brillouin zone for a period of 300 nm,

still in the weak coupling regime. For small incidence angles, only the silica-side

SPP is excited, the air-side SPP being excited for an incidence angle larger than

36.5° with an incidence wavelength between 500 and 630 nm.

To briefly summarize the agreement between the experimental and numerical

results,Tab. 3.1 sums up and compares both results for a grating periodicity of 300

nm and NP of 200 nm, illuminated under normal excitation, (see Figs. Fig. 3.8 (a)

, Fig. 3.10 (a) and Fig. 3.11 (a)). Numerical simulations highly confirm our experi-

mental results for all the different diameters and periodicities and the localized and

78


Figure 3.11: (a) Evolution with the period of the excitation wavelength of the two

PSP supported by the substrate: black: air-side PSP, red: silica-side PSP. The

incident planewave arrives normally to the interface, from the air. On the right

are plotted the profile of the electric field for different wavelengths indicated by

the numbers. (b) Folded dispersion curves of the two PSP modes associated to the

substrate, for a period of 300nm. The wavelength ❧ in vacuum varies from 500nm

to 900nm, k0=2♣/❧ and kx is the parallel component of the incident wavevector.

Dipolar mode

localized on

the upper surface of

the nanoparticle.

Gap modes (observed

as Fano profile

experimentally

because of the

coupling with SPP

Air-side SPP (m=1, n=0)

Experimental ∼520 nm 735 nm 569 nm

Numerical ∼520 nm 738 nm 560 nm

Table 3.1: Comparison of the numerical and experimental results of all the observed

plasmonic modes for nanoparticles having a diameter of 200 nm and a 300 nm

periodicity, for excitation in normal incidence.

delocalized modes are thus all well identified. The small discrepancies in resonant

wavelengths are probably due to minor fabrication imperfection both in the NP

preparation and the thin film deposition, such as size, thickness, roughness... This

is not the first time Fano resonances are reported in metal-dielectric-metal inter-

face. It was previously observed by Lodewijks et al [76], where it arised from the

coupling of dipole modes requiring relatively complicated NPs geometries. Up to

our knowledge, this is the first time a plasmonic Fano resonance with the narrow

localized mode being a gap mode is reported. The advantage is the tunability over

a wide wavelength range (600 to 800 nm) controlled by a simple variation of the

NP’s dimensions.

79

Chapter 3


interface

3.5.3 Angle resolved measurements

The optical setup which was used used was modified to tilt the sample, allowing

us to investigate illumination and collection angles beyond the traditional normal

incidence configuration, up to 60° under TM and TE polarization. After identifying

the nature of the localized and delocalized modes under normal incidence we be-

lieve angle-resolved measurements can highly enrich this study. The changes in the

excitation conditions induce wavelength shift of SPPs allowing a controlled overlap

and coupling with localized modes at specific angles. It also allows the observation

of new modes which do not appear in normal incidence, an example of which is

the air-side SPP that can be excited only at higher angles through folding of the

Brillouin zone as previously discussed and shown numerically in Fig. 3.11 .

Figure 3.12: . Evolution of extinction spectra versus illumination and collection

angles for a) TE and b) TM illumination.

Fig. 7.7 shows the evolution of the extinction spectra (-log(Itrans/Io), Io(resp.

Itrans) being the intensity transmitted with (resp. without) nanoparticles versus

angle for 200-nm diameter NPs with 300-nm periodicity. As expected and shown

on Fig. 3.11 (b), the ~560 nm air-side SPP mode red shifts with increasing angle

of incidence. These features are more prominent for the TM illumination since

SPPs are excited by TM illumination, however the same phenomenon can be seen

for TE polarization: here the diffraction process can create the vertical component

of the electric field necessary to excite SPPs, even from a TE incident wave. At

angles of 30°, a perfect overlap between the air SPP and the 520 nm LSP mode is

observed with a most intense unique resonance with a full width at half maximum

(FWHM) assessed to17 nm. Both the sharpness and the increase in intensity of this

resonance at 30° indicate a coupling and hybridization between the air-side SPP and

80


the LSP.Controlled coupling of localized and propagating modes leading to sharp

resonances could be very promising for sensing measurements and can increase the

sensitivity of such systems by one or two order of magnitudes.

Changes in the angles of illumination and collection induce no spectral shift for

gap modes. However the modifications in the excitation conditions alter both the

amplitude and phase of the gap mode and the broad continuum thus changing the

interference conditions. It is known that for Fano profiles even minor variations in

the strength or phase of the two different interfering modes can change the degree of

assymetry [123]. For this reason, we observe in our system different lineshapes for

the Fano resonance at different angles, where the degree of asymmetry decreases as

the angle increases, reaching an almost lorentzian profile for 60° with a wavelength

matching that of the gap mode on the single NP calculated previously (see Fig. 3.10

a).

The ON/OFF switch of the Fano resonance and the controlled overlap and

coupling between LSP and SPP lead to a fundamental understanding and control

over the parameters of MIM interfaces. This allows an exact engineering of similar

tunable systems favorable for numerous plasmonic and optical applications.

The angle-resolved measurements were supported by numerical simulations.

Fig. 3.13 (a) shows the angle-dependent extinction spectra computed with a periodic

version of the Green’s tensor method where the grating has a period of 300 nm, the

cylinder NPs are 200-nm large and 50-nm high. The distribution of the electric field

in the polarization plane for selected wavelengths and incidence angles is plotted

on the right of Fig. 3.13(a) (colorscale: electric field amplitude, green arrows: real

part of the electric field). The spectra are shown for TM incident polarization; the

incident magnetic field is then parallel to the substrate interfaces. A good agreement

is obtained with experiments where the localized and delocalized modes as well as

the experimental Fano profile are well reproduced.

In addition, new peaks (labeled 2 and 4) which appear for large incidence angle in

the measurements correspond to the excitation of other gap-modes not seen under

normal incidence due to their symmetry. Those as well as the gap mode observed for

normal incidence (3) are indicated by green lines on Fig. 3.13 (a). The corresponding

field distributions are plotted next to the spectra (labeled (1 to 6)). The air-side

SPP (labeled 5) is easily seen both in experiments and simulations at incidence

angles larger than 40°, however the silica-side SPP (labeled 6) is too narrow to be

resolved in the experiments. Overall, when comparing to the simulated peaks to the

experimental ones several conclusions can be drawn:

1) The experimental peaks are a bit wider compared to the simulated ones, which

explains that the number of resonances does not always correspond for every in-

cidence angle. This happens particularly when two peaks in the simulations are

very close to each other, like for example with the 50° incidence. For that value,

two modes are evidenced in the simulations near 720 nm, which are not resolved in

the experiment. One corresponds to the grating-excited silica-side SPP (710 nm),

81

Chapter 3


interface

Figure 3.13: (a) Extinction spectra of a 2D square grating of 3D gold nanoparticles

of 50 nm thickness and 200 nm diameter on the multilayered substrate, period

300nm and TM polarization. The solid black and red lines are eye guides for the

evolution of the SPP modes on air (black) and glass (red) side, the dashed lines

corresponds to extinction spectra for angles with 5° step between the main angles.

The vertical green lines indicate the position of the non-dispersive gap-modes.

The distribution of the electric field amplitude inside the polarization plane has

been plotted for few selected angles and wavelengths indicated in the spectra; the

green arrows are the real part of the electric field. (b) Comparison between the

extinction spectra for TM and TE excitation as a function of the incidence angle.

while the other corresponds to the Fano profile occurring at the gap-mode number

(labeled 3) (730 nm).

2) The higher angles peaks which seem to correspond to the same mode because

they have the same wavelength for different angles might differ in nature. Indeed,

as shown on Fig. 3.13 (a), the positions of the delocalized modes is extremely angle

sensitive, and their position is often very close to the one of the localized modes. For

example, the experimental spectra for 20° and 30° incidence angles show both one

mode at 655 nm. However its nature is different, and corresponds to the localized

mode (labeled 4) at 20° and to the substrate-side SPP for 30°, which appears more

intensely in the simulations while being close to the localized mode wavelength at

this angle.

3) Finally, we can see on Fig. 3.13 that the spectra for the TE polarization are less

angle-sensitive in the simulations than in the experiment. A major difference is

82


the intense mode which shows at 850 nm in the experiments for incidence angles

larger than 40°. The fact that it is angle dependent is an indication that it might

correspond to a delocalized mode, but it could not be reproduced in the simulations.

3.5.4 Understanding the Fano resonance

As we saw in the previous experimental section, two Fano resonances with

different lineshapes and probably different nature are observed in the MIM structure.

One is observed for the large NPs (diameter≥ 140 nm) with a linear red shift versus

the increase in diameter and the second is observed only for the 80 nm NPs at higher

wavelengths. Single NP simulations discussed previously showed that the localized

mode contributing to the Fano resonance is the gap mode for the larger NPs and

the longitudinal LSP for the 80 nm NP. However the question which remains is the

nature of the broad band coupling with the localized modes, and if it is the same for

the two Fano resonances. In a previous study, M. Svendhal and M. Kall [8] showed

how a Fano interference may result from the coupling of the localized LSP mode

of an Au cylinder with a continuum coming from the reflection of the broad white

ligtht at the interface. Since we evidenced the Fano resonances in an MIM structure,

it was reasonable to consider the possibility that the continuum coupling with the

discrete mode is the white light broad band transmission at the interface.

As we discussed previously (seeFig. 3.10 c) the amplitude of the field scattered

at infinity in the direction of the transmitted incident wave for a 200 nm NP and

80 nm NP. The big NPs shows an asymmetry similar to experiments in the profile,

while the 80 nm NPs look symmetrical. This implies that the Fano resonance for

the 80 nm NP could be due to the interaction of the LSP with the white light broad

band transmission at the interface. However this cannot be the case for the larger

NPs.

We propose in this part a simple analytical model developed by our colleague

Gaëtan Lévêque, which allows us to recover the different types of profile found

numerically and experimentally. The particle is modeled as a small polarizable

particle P(✇) with an effective polarizability α(✇), placed nearby a highly reflective

substrate described by its reflection and transmission coefficients R and T, as shown

in the schematic in Fig. 3.14. The emitting dipole is expressed as a function of the

incident plane wave as:

P (w) = α(w)Eloc = α(w)(Ei + Eref ) = (1 +R)α(w)Ei (3.2)

It is then very simple to express the specular reflected and transmitted components

of the electric field using the asymptotic expression of the Green’s tensor:

Eref,scatt(w) = (Gd∞ +Gs,ref

∞ )P (w) (3.3)

Eref,scatt(w) = Gs,trans∞ P (w) (3.4)

83

Chapter 3


interface

Figure 3.14: Schematic of a highly relective substrate, showing both the reflection

and transmission.

Indeed, if we forget the proportionality factor exp(ikR)/R, the components of the

direct, reflected and transmitted Green’s tensor reads in the normal direction:

Gd∞ = i

2, Gs,ref

∞ = iR2

, Gs,trans∞ = iT

2

Which gives:

Eref,scatt(w) =i

2(1 +R)P (w) =

iEi

2(1 +R)2α(w) (3.5)

Eref,scatt(w) =i

2TP (w) =

iEi

2T (1 +R)α(w) (3.6)

In the Green’s tensor expression, thei/2 factor is the phase retardation between the

plane wave at infinity and the radiating dipole P.

In the case considered here where the substrate is a mirror-like substrate, the

reflection coefficient R can be approximated by a small phase factorR ≈ −exp(−iε).In order to support this approximation. Fig. 3.15 shows the evolution with the

wavelength of the reflection coefficient on the experimental substrate with the 6nm-

silica/50nm-gold (dashed line) and for a semi-infinite gold substrate (solid line).

That approximation is very good for wavelengths larger than 600 nm. Using the

expansion R ≈ −1 + iε + (ε2)/2, the expression of the specular transmitted and

reflected fields reads then:

Eref,scatt(w) = iC

2(1 +R)2α(w) ≈ −iEi

C

2ε2α(w) (3.7)

Etrans,scatt(w) = iTC

2(1 +R)α(w) ≈ −EiT

C

2εα(w) (3.8)

84


Figure 3.15: Evolution with the wavelength of the reflection coefficient on the ex-

perimental substrate with the 6nm-silica/50nm-gold (dashed line) and for a semi-

infinite gold substrate (solid line).

An important difference between the reflected and the transmitted fields is their

phase (there amplitude are for both proportional to |❛|), which shows a shift by ♣/2.

This fact will have a huge impact on the field scattered by an ensemble of particles,

as the scattered field will interfere in a different way in reflection or in transmission

with the incident plane wave. Indeed, the total reflected and transmitted fields will

now read, with their corresponding amplitudes:

Eref (w) = REi + Eref,scatt(w) ≈ Ei

(R − i

C

2ε2α(w)

)(3.9)

which can be approximately reduced to:

Eref/Ei(w) ≈ 1 − 2Cε2Im(α)

Etrans(w) = TEi + Etrans,scatt(w) ≈ TEi

(1 − C

2εα(w)

)(3.10)

which also can be reduced to:

Etrans/Ei(w) ≈| T | (1 − CεRe(α)) (3.11)

Hence, the important result of this simple analytical model is that the change in

reflection is proportional to the imaginary part of the polarisability of the particle,

while the change in transmission is proportional to the real part of the polarisability.

85

Chapter 3


interface

Let us suppose now that the particle presents two localized resonances at wave-

lengths ❧1 and ❧2, the polarisability can then be written as:

α(w) = Ao +A1

w1 − w − iγ1

+A2

w2 − w − iγ2

(3.12)

with wi = 1λi

, and γi=∆λi

λi2

, the parameters were taken to match the 200 nm results:

| Ei |= 1 , Ao= 0, λ1 =1500 nm, ∆λ1= 150 nm, A1/γ1=7, λ2 =740 nm, ∆λ1=

30 nm, A2/γ2=1. Fig. 3.16 shows the extinction spectrum associated to the first,

long-wavelength resonance Im (α1(ω)) in blue, the extinction spectrum associated

to the second, short-wavelength resonance Im(α2(ω)) in red, and the extinction

spectrum associated with the bi-resonant particle Im(α(ω)) in black. As in that

model the particle can be considered as point-like, its extinction identifies itself to

its absorption, and we have a symmetric profile in the three considered cases. As

an indication is plotted as well the real part of the polarisaliblity.

Figure 3.16: the extinction spectrum associated to the first, long-wavelength res-

onance Im (α1(ω)) in blue, the extinction spectrum associated to the second,

short-wavelength resonance Im(α2(ω)) in red, and the extinction spectrum asso-

ciated with the bi-resonant particle Im(α(ω)) in black.

Next, we show in Fig. 3.17 the amplitude of the scattered fields, both proportional

to abs(α). As in the full Green’s tensor simulation, only the short wavelength mode

shows a Fano profile, due to its interference with the long-wavelength mode, which,

despite that its wavelength is twice the wavelength of the second mode, has still

a non-zero amplitude around λ2 due to its long tail. Everything happens as if the

long-wavelength mode behaves as a continuum near the short-wavelength resonance.

Finally, we show in Fig. 3.18 the total reflection and transmission in the spectular

directions, which is as expected totally symmetric in reflexion but both asymmetric

in transmission. Let us emphasize that the origin of the Fano is different for the long

86


Figure 3.17: a) the reflected and the transmitted scattered field b)the amplitude

of the scattered fields, both proportional to abs(❛). As in the full Green’s tensor

simulation.

and the short wavelength modes. For the long wavelength mode, the Fano results

from the interference of the incident transmitted wave with the field scattered by

the particle in transmission through the substrate, while for the short-wavelength

mode, the Fano results from the interference of that mode with the long tail of the

long-wavelength mode.

Figure 3.18: Theoretical reflection and transmission plots scaling as Im(α) and

Re(α) respectively.

As a conclusion, we do not need two modes in the localized system in order to

observe the Fano profile in transmission. Indeed, if the particle has only one mode

(A1=0), the Fano profile will occur due to the interference between the scattered

transmitted field and the incident transmitted wave. However, the system needs to

be bi-resonant in order to observe the Fano profile already in scattered field. In that

87

Chapter 3


interface

case, the asymmetry appears even if the broad mode is far from the narrow mode

as shown in the model, provided that its amplitude is non-negligible compared to

the amplitude of the narrow mode at its resonance wavelength.

In order to check the validity of this model we decided to perform reflection

measurements to check if the assymetric Fano profile observed in transmission is

symmetrical when measured in reflection. At this point unfortunately the 6 nm SiO2

sample we prepared earlier was no longer in good condition, therefore we prepared

a new sample, however ellipsometry measurements showed that the thickness of the

latter was 4.5 nm. Even though the thickness is not identical but this sample also

showed a similar Fano resonance when measured in transmission with a small shift

in the wavelength induced by the SiO2 thickness discrepancy. Although not ideal

(Since the fitting parameter used in the model was used based on a 6 nm spacer

system) but a comparison between transmission and reflection measurements on the

new sample can give a sufficient qualitative insight on the nature and symmetry of

the observed modes which can validate the suggested analytical model. Fig. 3.19

Figure 3.19: Experimental a) transmission and b) reflection measurements under

normal incidence for a 4.5 nm SiO2 MIM structure.

Shows Both a) transmission and b) reflection measurements, it is clear from these

results that the asymmetric Fano profile observed for big NPs when measured in

transmission has a symmetrical lorentzian shape when measured in reflection. These

experimental results shows similar behaviour compared to those obtained using the

developed model, which validates the model and confirms the origin of the Fano as

the interference between the broad dipolar mode at high wavelengths 1500 nm and

the narrow mode at 740nm.

3.6 Conclusion

In this chapter, we thoroughly investigated the optical properties of MIM structure

with a thin 6 nm SiO2 spacer layer. We identified the plasmonic modes through a

88

3.6 Conclusion

cross comparison between the experimental and theoretical results, and we discussed

their origin. Further measurements with different angle of incident showed the high

dependency of these modes on the illumination conditions and the possibility to

couple and decouple localized and delocalized modes. Finally we developed an

anlytical model to undertand and explain the origin of the Fano profiles observed.

89

4 Influence of the spacer thicknesson the plasmonic properties

4.1 Introduction

We have studied in the previous chapter coupled NP/film systems and evidenced

their rich plasmonic properties. We identified the different localized and delocalized

resonances and their dependency on the size, spacing and angle of incidence. In

this chapter, we aim at studying the evolution of the different plasmonic modes as

the thickness increases. As we discussed earlier, separate groups explored different

spacer thicknesses varying from 0 [33, 42] to 20 nm [32, 34] up to 50 nm [33].

However no consistent and systematic work was done to study solely the influence

of the thickness on the plasmonic modes. This motivated us to perform a full

comprehensive study on coupled Au NPs/Au film systems with SiO2 spacer layers

of different thicknesses varying from 10 up to 100 nm. In this chapter, we will

discuss the effect of the thickness on the resonances, where the plasmonic modes

broaden and shift or even disappear as the thickness increases. We will also discuss

the potential of MIM structures with thick spacers as super absorber.

4.2 Sample preparation and structural

characterization

The samples were prepared similarly to those discussed in the previous chapter.

To prepare the substrate, a 50 nm Au film was deposited on a clean glass substrate

via PVD deposition, followed by the deposition of a SiO2 layer with a specific thick-

ness. On top of the different substrates, identical NPs with diameters varying from

80 nm up to 200 nm, a center to center distance of 300 or 450 nm, and a height of

50 nm were deposited using the EBL technique. For all the samples, we used a 3

nm Cr adhesion layer to avoid undesired adhesion problems between Au NPs and

the SiO2 film.

Measuring the exact thickness of the SiO2 layer is very critical for this study,

especially because the thin films prepared with PVD might show discrepancies be-

tween the expected and the real value of the thickness as well as some inhomogeneity.

For this reason, we measured using an ellipsometer the thickness of the SiO2 layer

91

Chapter 4 Influence of the spacer thickness on the plasmonic properties

on all the prepared samples and for different locations on each sample. We prepared

a set of 8 samples with SiO2 layer thickness of: 10, 20, 30, 40, 50, 60, 80, and 100

nm. Ellipsometry measurements demonstrated that the error in the thickness can

go up to ± 5 nm. To give a more detailed description of the spacer layer thickness

we will list in Tab. 4.1 the measured thickness by the ellipsometer for the differ-

ent substrates compared to the expected ones. Such a discrepancy can be high for

small thicknesses. However as the thickness increases, it becomes acceptable and

the optical properties can still be investigated.

Expected thickness (nm) 10 20 30 40 50 60 80 100

Measured thickness (nm) 8 24 27 44 51 64 75 98

Table 4.1: Comparison between the expected and the measured values of the SiO2

spacer layer thickness.

As for the NPs we regularly performed SEM images to ensure that the size

and shape of the NPs are within the acceptable range. The average of the error in

the diameter is around ± 2 nm and the center to center distance is ± 1 nm.

4.3 Evolution of Plasmonic modes versus the increase

in the spacer thickness

In order to study the evolution of the plasmonic modes for different spacer

thickness, we perform both experimental and numerical measurements. For the

experimental measurements, we perform UV-VIS spectroscopy (using the NIKON

spectrometer described in Chapter 2) to measure the extinction spectra. We measure

the extinction of the NPs on a reflective substrate and not in a solution which makes

the exact definition of extinction for such a system somehow delicate. To avoid

any ambiguity, the extinction in our measurements corresponds to −log(T/To) ,

where T is the optical transmission measured on regions with NPs and To the

transmission measured on regions without NPs. The same quantity was measured

in the simulations performed by our colleague Gaëtan Lévêque using the Green’s

tensor method [125]. The different spectra for the different thicknesses are presented

on the same graph, and separated by a constant factor on the y-axis for a better

representation of the data. The measurements for NPs with a periodicity of 300

nm and 4 different diameters: 80, 110, 140, and 200 nm are presented. Since the

behavior of small NPs (diameter ≤ 110 nm) is different from that of larger ones and

thus we will discuss their results in separate sub-sections for the sake of clarity. To

make the comparison easier, we used identical colors for experimental and numerical

spectra of the different spacers for all the different NPs.

92

4.3 Evolution of Plasmonic modes versus the increase in the spacer thickness

4.3.1 Small NPs (diameter ≤ 110 nm)

Fig. 4.1 and Fig. 4.2 show in part a) the experimental and part b) numerical ex-

tinction spectra for the different spacers for the 80 nm NPs and the 110 nm NPs

respectively.

Figure 4.1: a) Experimental and b) numerical extinction spectra for NPs grating

with a diameter of 80 nm and a periodicity of 300 nm, deposited on Au film with

an SiO2 spacer layer with thickness varying from 10 up to 100 nm.

We have shown in the previous chapter that MIM interfaces could be fairly com-

plex plasmonic structures, holding both delocalized and localized modes. Since the

several modes in these systems may present different origins, their response versus

the change in the dielectric spacer thickness can also strongly vary. Therefore, the

different modes will be presented in separate subsections. The main modes which

can be observed are:

1) a low wavelength LSP mode ∼ 520 nm,

2) a delocalized glass SPP mode ∼ 560 nm,

3) a longitudinal LSP mode whose resonance at higher wavelengths.

520 nm mode The observation of a LSP resonance at such a low wavelength was

relatively new to us. However we established a clear definition of this mode in the

previous chapter. This mode can be defined as a LSP whose hot spots are pushed to

93





the top of the NPs due to a mirror effect played by the Au film. We have shown that

this mode is period independent which confirms that it is a localized mode and we

have also shown that it is almost independent of the NP size which is somehow less

expected for LSP resonances. However one important note that we did not discuss

previously is that the intensity of this mode increases as the diameter increase. For

the 80 and 110 nm NPs, we can clearly see this mode around 520 nm for a small

thickness of 10 nm, as the thickness increases the identification of this mode becomes

harder. However if we plot each thickness separately and we zoom in more around

520 nm we can still detect up to a spacer thickness of 50 nm a small bump which we

believe is the 520 nm LSP mode. While for the bigger NPs the identification of this

mode is easier. The reason why the intensity of this mode drastically diminishes

as the thickness of the dielectric increases can be attributed to three effects: First

the efficiency of the Au film as a mirror decreases as the spacer thickness increases

which affects the intensity of the resonance. Second, this mode is spectrally close to

the SPP mode (∼560 nm) which makes it hard to identify it, since the two modes

can be seen as a single broad peak. Finally, for bigger thicknesses beyond 50 nm,

the longitudinal LSP shifts to the blue and its resonance position becomes close

to the 520 nm mode. The longitudinal LSP is an intense mode which makes the

identification of the 520 nm mode almost impossible. A final note regarding this

94


mode is that its resonant wavelength is independent of the thickness of the dielectric,

this is probably because of the nature of this mode with the hot spots on nthe top

corners, i.e not in contact with the dielectric.

Glass SPP We showed in the previous chapter that for a small spacer thickness a

glass SPP can be excited under normal incidence. For a grating periodicity of 300

nm, the resonance wavelength of this SPP is 560 nm as we saw in chapter 3 (Fig. 3.8

a) and for a spacer thickness of 6 nm. On the other hand simulations performed

by our colleague Gaëtan Lévêque showed that the position of this glass SPP was

not very sensitive to the change in the dielectric spacer thickness. The resonant

wavelength was measured theoretically for the 25, 50 and 100 nm spacers and the

resonant positions were 564, 570, and 580 nm respectively. By looking at the spectra

in Fig. 4.1 and Fig. 4.2 we can see that a resonance around 560 nm is present for

all the spacer thicknesses and shifts with only few nms as the thickness increases.

However for thick spacers (≥ 60 nm), this resonance becomes less clearly defined.

At these big thicknesses, the small red shift of the glass SPP and the blue shift of the

longitudinal LSP result in an overlap between these two, and only a single peak can

be experimentally resolved. However numerically the resonance can still be resolved

up to a thickness of 80 nm. This can be explained by the sharp edges of the NPs

in simulations compared to experiments, and to the NA used for illumination in

the experiments. Even with these small differences, the experimental and numerical

results can be considered in very good agreement.

Longitudinal LSP mode The longitudinal LSP mode is a typical mode of nanocylin-

ders, exhibiting a red shift of the LSPR with the increase of the NP diameter as we

showed in the section 2.3. We also evidenced that for NPs with diameters varying

from 80 up to 200 nm fabricated directly on a quartz substrate, the longitudinal

LSP resonance lies in a wavelength window between 600 and 700 nm depending on

the NPs size. When an Au film is added on top of the quartz substrate, we can

no longer measure this longitudinal LSP mode since it is red shifted beyond the

limit of our spectrometer. Even though gold has a low index of refraction at the

visible wavelength, the LSP mode shifts to the red when the NPs are in contact

with Au film. The reason why we observe this shift was discussed in chapter 3,

and is due to the LSP dipole mode interaction with its image dipole in the mirror

through its evanescent near-field which thus shifts the resonance to higher wave-

lengths. The evanescent field of a particle is size dependent, where the decay is in

first approximation equal to D/2π, with D the NP diameter.

For a small spacer thicknesses of 10 nm, the longitudinal LSP resonance of the

80 nm NP is seen around 790 nm experimentally (see Fig. 4.1 (a)) and at 700 nm

theoretically (see Fig. 4.1 b) while for the 110 nm NP, this resonance could not be

measured experimentally (see Fig. 4.2 (a)) and is seen at 920 nm theoretically (see

Fig. 4.2 (b)). The reason for this discrepancy between the experimental and the

95


theoretical results comes from the experimental thickness of the spacer layer, which

based on ellipsometry measurements was not exactly equal to 10 nm. This indicates

that the position of the LSP mode is very sensitive to the exact thickness especially

for thin spacers. To confirm this, we compare the resonance for the 80 nm NP at 790

nm for the 8 nm spacer with the resonance at 840 nm for the 6 nm spacer studied

in the previous chapter; where only a 2 nm difference in the thickness induces a

50 nm shift in the resonance. As the spacer thickness increases up to 30 nm the

longitudinal LSP exhibits a fast blue shift. For example, in the case of the 80 nm

NP, the resonance blue shifts from 790 nm for a spacer of 10 nm to 600 nm for a

spacer of 30 nm. Beyond this spacer thickness, the resonance reaches a threshold

and shifts only by few nms for any further increase in the thickness. For the thickest

spacer of 80 and 100 nm, a slight red shift is still observed. This red shift is the

result of the overlap of the longitudinal LSP with the glass SPP which resonates at

higher wavelengths for these thick spacers.

We also note that for all the thicknesses larger than 20 nm, there is a good

agreement between the experimental and numerical results since the small discrep-

ancies in the experimental thicknesses of the spacers become less important. As the

spacer thickness increases, the value of the effective index of refraction (neff ) also

increases because of the high index of refraction of SiO2. As neff increases, a red

shift in the LSP should be expected. However as we discussed in chapter 1 and 2 it

is not only the index of refraction of the surrounding media which affects the reso-

nance but also the interaction of the LSP dipole mode with its image in the mirror.

In order to account for the film dielectric properties in this interaction the decay

becomes D/2π*N, with N being the refraction index of the spacer. The strength of

the interaction of the LSP mode with its image is highly dependent on the separa-

tion distance between the NPs and the film. Based on the classical F..orster model,

the interaction between two dipoles depends on the distance separating the accep-

tor and the donor with a d−6 dependency [126]. This explains the huge blue shifts

recorded for small variations of the spacer thickness for small spacers (less than 30

nm). Beyond this thickness, the interaction is already weakened and thus no longer

affects the LSP resonance. Observing the evolution of this mode versus the increase

in the spacer thickness, it demonstrates that the shift exhibit an exponential decay

versus the thickness.

4.3.2 Large NPs (diameters≥140 nm)

We also performed extinction spectroscopy on larger NPs of 140 and 200 nm

diameters. Fig. 4.3 and Fig. 4.4 compare the experimental (see part a) and the

numerical spectra (see part b) for the different thicknesses. Similarly to the smaller

NPs, the 520 nm mode can still be observed and a gap mode up to a certain thickness

which was not observed earlier as well as the glass SPP around 560 nm, and the

longitudinal LSP mode. We will briefly discuss the behavior of the modes already

observed for the small NPs, while focusing more on the gap mode since it was not

96


observed for the small NPs and the longitudinal LSP mode because its behavior

differs from that of small NPs with less agreement between the experimental and

numerical results.




520 nm mode In the case of no spacer layer, the intensity of the 520 nm mode

increases as the NPs size increases. Thus we notice that when a spacer layer is

introduced and its thickness is gradually increased, it is easier to identify the 520

nm mode for the large NPs (≥140 nm) compared to smaller ones. However for spacer

layer thicknesses larger than 50 nm, the 520 nm mode can no longer be resolved even

for the biggest NPs of 200 nm diameter (see Fig. 4.4).

Glass SPP The behavior of this mode for big NPs is very similar to the smaller

ones. Indeed, the resonant wavelength of this grating coupled SPP depends on the

center to center distance and not on the size of the NPs and thus remains around

560-580 nm. The only difference is that this mode is more clearly identified for

large NPs and a distinct peak can be observed up to a spacer thickness of 80 nm.

For the 100 nm spacer, this mode can no longer be resolved since it overlaps with

the longitudinal LSP. Experiments and simulations show a good agreement in the

resonance wavelength and profile for the different spacers.

97





Gap mode In the previous chapter, we saw that for a spacer thickness of 6 nm

a Fano resonance can be excited only for NPs larger than 140 nm in a wavelength

window of 650 to 750 nm. By plotting the electric field map, we showed that

the localized mode interacting with the continuum in this Fano resonance is a gap

mode localized between the NPs and the film. This gap mode originates from the

fact that the area of the spacer under the NP behaves like a resonant cavity for

the propagative surface plasmon modes. Indeed, the incidence planewave excites

by diffraction the SPP waves which are bounded to the 50 nm metal film on the

top interface. This SPP will then excite a delocalized metal-insulator-metal SPP

(MIM-SPP) under the particle which, due to the strong impedance mismatch at

the NP edges, will bounce back and forth under the cylinder and form resonant

patterns for specific wavelengths. In Fig. 4.4 we see that this gap mode can still be

excited up to thicknesses of 10 nm. For thicker spacer the distance under the NP

no longer acts like a resonant cavity and no gap modes can be excited. For the 10

nm spacer, the gap mode exhibits a typical Lorentzian symmetrical lineshape which

indicates that it does not interact with a broad band continuum like for the 6 nm

spacer. The interaction conditions are very sensitive to the spacer layer thickness

where even few nm difference can change the resonance from asymmetric Fano to

a symmetric resonance. A comparison between the experimental and numerical

simulations for the 10 nm spacer (black line in Fig. 4.3 and Fig. 4.4) show that there

98

4.4 Effect of the Numerical Aperture

is a disagreement in the position of the gap mode. Again this can be explained by

the difference between the real and the expected thickness. The resonant position

of the gap mode red shifts as the diameter of the NP increases with the 140 and 200

nm NPs resonances at 629 and 690 nm respectively.

Longitudinal LSP mode Similarly to small NPs the longitudinal LSP mode blue

shifts as the dielectric spacer increases. For large NPs, the blue shift remains re-

markable up to thicknesses of 40 nm compared to 30 nm for the small NPs. For the

200 nm NPs we can only observe this mode for thicknesses of 20 nm and above, since

for all smaller spacers the mode is in the red beyond the range of our spectrometer.

Indeed, simulations demonstrates that the resonance wavelength is even higher than

1000 nm. For thicker spacers, there are much bigger differences between the exper-

iment and the numerical simulations as shown in Fig. 4.3 and Fig. 4.4. It is very

clear that when the spacer thickness increases, the simulation predicts rapid blue

shift of the extinction maxima while in the experiment a large maximum appears in

the red around 700-750 nm, and is particularly pronounced for the 100-nm spacer.

These discrepancies might find their origin in the uncertainty of fabrication, more

particularly in the size and shape of the particle and/or the spacer thickness. Nev-

ertheless, since the fabrication process is well-controlled, this alone cannot explain

the differences between the computed and measured spectra. However, in order to

collect more signal experimentally, a rather large numerical aperture of about 0.45

has been employed in the illumination which means that the sample is placed under

a range of plane waves with incidence angles between 0° and 26°. This is rather far

from the single plane wave illumination employed in the numerical simulation. In

the following section, we will therefore discuss the effect of the NA on the plasmonic

modes.


We showed in the previous chapter that the illumination conditions can highly

affect the plasmonic resonances. Angle resolved measurements showed the possibility

to excite SPP modes which could not be excited under normal incidence like the air

SPP, and to induce shifts in other plasmonic modes. Throughout this chapter, we

considered only normal illumination both experimentally and numerically. However

experimentally, we are bound by a NA of illumination equal to 0.45 meaning that

the sample is placed under a range of plane waves with incidence angles between 0°

and 26°, which is rather far from the single plane wave illumination employed in the

numerical simulation. In order to verify this fact, we have performed simulations

for several values of θ and ϕ, and averaged the transmitted intensities as follow:

Extinction = −log TT o

(4.1)

99


with:

T =

π/2ˆ

0

dθw(θ)

ˆ 2π

0

dϕT (θ, ϕ) (4.2)

and:

T0 =

ˆ π/2

0

dθw(θ)

ˆ 2π

0

dϕT0(θ, ϕ) (4.3)

where T(θ, ϕ) is the transmission for a plane wave incident along the (θ, ϕ) direction

through the full system, T0(θ, ϕ) is the transmission for a plane wave incident along

the (θ, ϕ) direction through the substrate without the nano-cylinders, and w(θ) a

weight function only dependent on θ. Moreover, the incident field has been polar-

ized by taking the electric field amplitude in the (θ, ϕ) direction having the form

Exi+Ezk for p polarization (no y component), and Eyj+Ezk for s polarization (no

x component). To simplify the numerical averaging process and make it faster, we

have computed the transmissions for θ = 0°, 5°, 10°, . . . 50° and for ϕ = 0°, 22.5°,

45°, 67.5° . . . making use of symmetries between angles and polarization. For exam-

ple, results are the same for (ϕ=0°,p) and (ϕ=90°,s), or (ϕ=45°,p) and (ϕ=45°,s).

Fig. 4.5 shows the experimental (see part a) and numerical (see part b) spectra for

Figure 4.5: a) Experimental and b) Numerical (with an angle average to account

for the NA) extinction spectra for NPs of different diameters and a periodicity of

300 nm on an MIM structure with 100 nm SiO2spacer layer.

NPs of different diameters on a 100 nm SiO2 spacer. On the y-axis, the spectra were

separated by a constant factor for a better visual representation. These numerical

results illustrate the averaging process where w has the following form:

w(θ) =1

1 + exp((θ − θs)/ws)(4.4)

with θs= 20° and ws= 5°. We can see that the experimental shape of the extinction

is very well reproduced, together with its main features: a minimum at 540 nm, an

100


asymmetry of the main resonance, an increase of the extinction below 500 nm with

the diameter, and a change in the concavity of the red tail of the main resonance

when the diameter increases. In order to explain the origin of the differences, we

then present the evolution of the extinction spectra with θ, for different fixed values

of ϕ, with a spacer equal to 100 nm (see Fig. 4.6). The extinction spectra are clearly

Figure 4.6: Numerical simulations showing the evolution of the extinction spectra

with θ, for different fixed values of ϕ, for a spacer thickness of 100 nm.

very ϕ-dependent. In particular, it seems that the broad maximum around 750

nm on Fig. 4.5(b) comes from the off axis (ϕ=22.5°-67.5°) illuminations. We can

notice as well a lot of fast moving peaks between 500 nm and 700 nm which comes

from the folding of the SPP dispersion curve in the first Brillouin zone. Hence, the

numerical aperture averages all these sharp resonances in the measured extinction,

which results in a large broadening of the theoretical resonance computed from a

single plane wave in normal incidence. As a conclusion, we here demonstrate that for

large spacers the extinction spectra change dramatically around normal incidence,

apparently much more than for thin spacers.

101


4.5 Narrow band super absorber

High light absorption is an essential optical phenomena necessary for various

applications. Historically semiconductors served as the first candidates for these

systems with their well defined absorption bands. However super absorbers based on

semi-conductors are very challenging, mainly due to the limitation in the absorption

band tuning. On the other hand, plasmonic based metamaterials offers promising

potential in light absorption, with a spectrally tunable bandwidth based on the

tunable optical resonance. Investigation of coupled NP/film structures had its early

start in the late 80’s with Holland et Hall [93, 92], following that several studies on

different configurations of these systems were conducted [127, 31, 34, 86, 128, 129].

However the first observation of high absorbing systems based on coupled NP/film

structures was reported by Papanikolaou et al [86], even though high absorption was

not the aim of their study. The first well established study on super absorbers based

on meta-materials was done by Padila et al [130], actually super absorbers based

on electric dipole-film interaction started with them. After the work of Padila et

al, several experimental studies investigated the MIM structures for super absorber

applications [9, 100, 10, 131]. In most of these studies, the high absorption originates

from the collective effort of the dipolar resonance and its interaction with the metallic

film. The optical resonance indeed leads to an increase in the electric field and thus

increases the absorption cross section. This enhancement and the lossy nature of

metals result in strong absorption spectral features.

The main objective of this thesis was to study the optical properties of coupled

NP/film systems. This in-depth study can be very useful for future designs of super

absorbers based on MIM structures, especially with the high tunability and the

possibility to easily overlap and/or couple the localized and delocalized plasmonic

modes as we showed earlier. The coupled NP/film system with 100 nm thick SiO2

spacer shows appealing high absorption features. For the sample with a thick spacer,

the red shift of the glass SPP to 580 nm and the blue shift of the longitudinal LSP

lead to an overlap of the two, which results in an increase in the absorption efficiency.

Fig. 4.7(a) shows the extinction spectra under TM polarization performed on our

home made UV-VIS setup. For the different NPs deposited on the 100 nm spacer

sample, as the particles size increases the absorption also increases. This can be

explained by the pronounced intensity of both the SPP and the LSP mode of the

system for larger NPs. The different absorption percentages for the different NPs

are presented in Tab. 4.2. The absorption is calculated using the formula A = 1−T ,

where A is the absorbance and T the transmission. Large NPs show high potential

for super absorber application, with a 94 % absorption at a wavelength of 760 to

790 nm. As the NPs size decrease the percentage drops down to 38 % due to the

reduced intensity of the LSP and SPP mode.

To confirm our assumption that the origin of the high absorption comes from the

overlap of localized and delocalized plasmonic modes, we performed measurements

on the same substrate but for NPs with a grating periodicity of 450 nm. Changing

102

4.5 Narrow band super absorber

Figure 4.7: Experimental extinction spectra under TM polarization for NPs of

different diameters and a) 300 nm and b) 450 nm periodicity.

NPs diameter (nm) 80 110 140 170 200

Absorption percentage (p=300 nm) 38 71 91.6 93.3 94

Absorption percentage (p=450 nm) - 46 76.5 84 84.86

Table 4.2: Absorption percentage for NPs of different diameters on 100 nm SiO2/50

nm au film substrate.

the grating periodicity leads to a change in the SPP resonant wavelength and thus

the overlap between this mode and the LSP mode might no longer exist. Fig. 4.7 (b)

shows the extinction spectra for the 450 nm periodicity. Both an air-side SPP and

a glass SPP are excited as well as a longitudinal LSP. However no overlap between

the three exist. The absorption percentage for the biggest NPs of 200 nm drops

down to 84% when there is no overlap. From these results, we can conclude that

the longitudinal LSP mode in this thick spacer MIM structure is itself appealing for

high absorption applications. However an overlap of this mode with the delocalized

SPP can result in a high efficiency super absorber. Even with the extremely high

absorbance of this interface, two main limitations have to be faced: First the ab-

sorbance is only over a narrow wavelength band, but this can have its advantages

especially for optical filters application. Second it is a lithography based technique

which makes it harder to integrate in commercial application. Nevertheless the com-

pact size of the structures can be easily used in photonic devices. High absorption

was not an objective of this work,Nevertheless with these promising results, it is

very interesting to pursue the investigation on super absorbers based on MIM inter-

faces. A lot of work can still be done to optimize the absorption of these systems,

which paves the way towards well controlled and engineered super absorber MIM

interfaces.

103


4.6 Conclusion

The systematic study of MIM structures with different insulator layer thick-

nesses resulted in a deeper understanding of the optical properties of these systems.

We identified the different localized and delocalized plasmonic modes and their evo-

lution as the spacer layer thickness increases. A cross comparison between the ex-

perimental and numerical results allowed for deeper understanding of these systems

and their high possible potential for different applications. We also evidenced the

possibility to produce high narrow band absorbers with an absorption % up to 94

with thick spacer MIM structures, by the overlap of localized and delocalized plas-

monic modes. The different measurements and conclusions drawn from this study

is highly important for future engineering of MIM systems to meet the requirements

of several optical applications at different wavelengths in the visible range.

104

5 Graphene a sub-nanometer twodimensional spacer layer

5.1 Introduction

Research on carbon based materials such as fullerene and carbon nanotubes

stems back to 30 years with the first discovery of fullerenes in 1985 [132]. However

theoretical works dating back to 1947 suggested that an isolated 2D layer of graphite

or what is known as graphene could be the ultimate carbon based material with ex-

traordinary properties such as relativistic behavior of electrons [133]. Producing a

free-standing form of planar graphene was assumed experimentally impossible, un-

til 2004 when Andre Geim, Kostya Novoselov and coworkers from the University

of Manchester in the United Kingdom obtained a 2D graphene layer by cleaving

a graphite sample with a sticky tape: it marked the beginning of the graphene

“gold rush”. Graphene is a single-layer honeycomb lattice structure with only 0.3

nm thickness, considered the super nanomaterial nowadays due to its remarkable

electrical, mechanical, optical, and thermal properties. It is chemically stable, with

a high specific surface area, high optical transmittance even higher than ITO, high

elasticity, tunable band gap, and it can be easily chemically functionalized which al-

lows tuning its properties [134, 135]. Theoretically to determine graphene properties

it can be considered as a 2D gas of charged carriers with high mobility. The elec-

trons in graphene are massless due to their unconventional collective behavior and

are thus considered as Dirac fermions which allows investigating strange relativis-

tic effects. Graphene can be currently synthesized either through micromechanical

exfoliation using a tape or through chemical vapor deposition (CVD). The first tech-

nique provides higher quality graphene with less impurities but is limited in size to

few mms, while the second is more practical for large scale applications [136]. All

the extraordinary properties of graphene makes it appealing both for fundamen-

tal studies and applications especially in photonics, plasmonics and optoelectronics

[133, 134]. Typical plasmonic materials are usually metals more specifically Au and

Ag with their plasmonic modes in the visible range. With the charge carriers densi-

ties achievable in graphene, it can exhibit plamonic properties in the infrared region

[137]. All the properties of plasmonics in the visible range can be transmitted to

the infrared and applied to graphene which makes it complementary to metals and

increases the wavelength band of plasmonic applications. We have seen in the pre-

vious chapters how the LSPR behavior of metallic nanostructures on the top of a

105

Chapter 5 Graphene a sub-nanometer two dimensional spacer layer

metal thin film is exquisitely sensitive to the spacer distance of the film to the NPs.

This motivated us to investigate the influence of using few-layered graphene spacer

to go do down to subnanometer seperations. The interest of graphene for plasmonic

devices has already been highlighted in several recent papers [138, 139, 140]. Sev-

eral groups investigated the potential of graphene as an alternative coating layer for

silver [141, 142] and gold [143] based surface plasmon resonance (SPR). Adopting

graphene to these systems can present several advantages:

1) the high volume to surface ratio of graphene is expected to enhance the efficient

adsorption of biomolecules compared to naked gold,

2) the carbon-based ring structure allows π stacking interaction with the hexago-

nal cells of graphene which is supposed to increase the adsorption of organic and

biological molecules,

3) controlling the thickness of the spacer through controlling the number of graphene

layers transferred onto the metal interface allows tuning the SPR response and the

sensitivity of SPR measurements [143].

While potential of graphene coated SPR sensors was exploited, no research

on LSPR behavior of metallic NPs on the top of a metal thin film with graphene

sandwiched in between has been done. In the second part of this research work we

take advantage of the ultra thin (0.34 nm) 2D structure of graphene [144] as a high

optical index non dielectric spacer between a flat Au film and Au NP array and we

investigate its influence on the plasmonic behavior of the newly designed interface.

To better understand this graphene based plasmonic system we compare it to a

similar one without the graphene layer previously investigated in chapter 3. Details

on the sample preperation and structural and optical properties are discussed in the

following sections. We also investigate the interaction of graphene with metal and

the dependence and strength of this interaction on the number of graphene layers.

And finally we study the possible doping of graphene by metallic NPs deposited on

it and how this doping tunes the plasmonic properties of such system.

5.2 Understanding the Raman spectra of graphene

With the fast rise of graphene, Raman spectroscopy became an essential com-

mon tool used for its characterization. Raman spectra indeed provides a wealth of

information on graphene such as the number of layers and thickness, doping, thermal

conductivity and strain disorder. The 2 main peaks of a graphene Raman spectra are

the G band and the 2D band (see Fig. 5.1). The G band corresponds to the in-plane

vibrational mode of the sp2 hybridized carbon atoms and it appears around 1580

cm−1. This band is highly sensitive to strain effects and to the number of graphene

layers. When the number of graphene layers increases, the G band frequency blue

shifts due to the softening in the bonds induced by the addition of a graphene layer

106

5.2 Understanding the Raman spectra of graphene

[144]. As for the 2D band it is a two phonon vibrational process, this band is res-

onant and thus exhibits a dipersive behavior (i.e it depends on the wavelength of

the laser excitation). Its position is around 2700 cm−1. The 2D band also gives an

indication on the number of graphene layers. However it is more complex as the

shape, position, and intensity of this band are all sensitive to the number of layers.

For monolayer graphene, the 2D band is a single sharp symmetric mode which can

be fitted with a single lorentzian. As the number of layers increases, this mode

is splitted and four lorentzians are required to fit the peak. This corresponds to

the four possiblities of double resonance scattering processes compared to the single

possibility in case of monolayer. A third peak that is not observed for a defect free

graphene layer is the D peak, for this reason it is called the disorder band. Similar

to the 2D peak, it is also a dispersive band observed between 1270 and 1459 cm−1

depending on the excitation wavelength. It represents a ring breathing mode from

sp2 carbon rings and it can only be activated when adjacent to a graphene edge or

a defect. Another defect peak can be observed for graphene which is a splitting of

the G peak and is called the D’ peak at 1620 cm−1. This peak is observed when the

localized vibrational modes of the impurities and surface charges interact with the

extended phonon modes of graphene. Fig. 5.1 shows a typical Raman spectra of a

pristine single layer graphene and a graphene layer with defects showing both the

typical graphene bands and the defect bands.

Figure 5.1: Raman spectra of a) perfect graphene monolayer, and b) disordered

graphene layer.

Besides the information provided by the different Raman bands, the ratio of the

2D band and the D band to the G band can also be very informative. The I2D/IG

ratio can be used to determine both the number of graphene layers and doping.

As the number of graphene layers increases, the intensity of I2D/IG is expected to

decrease, also the intensity decreases when the graphene is doped. However it is

more complicated to confirm graphene doping and one should also examine the

shifts in the 2D and G bands. Graphene doping leads to a shift in the Fermi level

away from the Dirac point. This decreases the probability of excited charge carrier

107


recombination [145] which reduces the phonon energy of the G peak resulting in

an increase in the frequency [146]. On the other hand, an increase in the electron

concentration expands the crystal lattice which decreases the energy of the Raman

phonons and results in a decrease in the 2D band position [146] and intensity. In

summary, as the I2D/IG decreases the doping level increases, an upshift of both the

G band and the 2D band indicates p-doping of graphene, while an upshift of the

G band and a downshift of the 2D band indicates n-doping of graphene. As for

the ID/IG ratio it is an indication of the defect density: it increases with increased

defects as a higher defect density creates more elastic scattering.

5.3 Sample Preparation

In this work, we investigated different graphene interfaces: metallic NP/film system

with graphene as a spacer layer, graphene coated SPR surfaces, and graphene on

quartz substrates with Au and Ag NPs deposited on top. To prepare the first two

kinds of samples, we had to transfer graphene nanosheets onto SPR surfaces (i.e a 50

nm Au film coated glass substrate). This was done using the dry transfer technique

[147, 29]. The graphene transfer is achieved using a commercial thermal release tape

which enables the mechanical transfer of 1 mm2 graphene sheet originally grown by

chemical vapor deposition (CVD) on nickel-coated silicon substrate onto classical

SPR interface (glass/5 nm Ti/ 50 nm Au). After the transfer, the tape is removed

by annealing the interface at 120 °C for 2-3 min and then the interface is rinsed

with acetone. A final step is required to remove all the tape residue and a high

temperature annealing at 500 °C for 360 min is performed. Fig. 5.2 demonstrates

the steps required to transfer the graphene sheet. Once graphene is coated on the

SPR chip, Au NPs can be fabricated using EBL on top of graphene to form the

NP/film system. The designed NPs are 50 nm high with a center to center distance

of 300 nm and diameter of: 50, 80, 110, and 140 nm. As for the samples with

monolayer graphene on quartz they were bought from ACS materials, a company

which provides good quality CVD graphene on different substrates. Arrays of Au

and Ag NPs with a periodicity of 300 nm and a set of diameters of 140, 170 and

200 nm were deposited using the EBL technique on the monolayer graphene/quartz

interface.


It is essential to control the quality of graphene sheets before and after the

transfer process, to make sure the transferred graphene is still of good quality as

both mechanical and chemical damages can occur during the process. To do so

we performed Raman spectroscopy on graphene grown on nickel foil and graphene

108


Figure 5.2: Schematic illustration of the fabrication steps of a graphene-based SPR

interfaces decorated with an ordered array of gold NPs using Electron Beam

Lithography.

transferred onto gold film. We also performed XPS spectroscopy to test the chem-

ical state of the graphene sheets. Raman spectroscopy and X-Ray photoelectron

spectroscopy (XPS) are adopted techniques to verify if the transfer process occurs

without oxidation or carbonization. Fig. 5.3 (part a) presents the Raman spectra

of the transferred graphene: the D-band, G-band and 2D-band Raman signals are

clearly seen both on nickel foil and gold. The intensity of the D-band is very small

for graphene formed by CVD on nickel foil, indicating a high quality graphene layer.

As for the XPS measurements Fig. 5.3 (part b) presents the integrated intensity ra-

tio of the D peak to the G peak (I2D/IG) of the Raman spectra increases from 0.52

before transfer to 0.75 after transfer, suggesting that some unavoidable mechanical

defects do result from the transfer process. The C1s core level high resolution XPS

spectrum (see Fig. 5.3 b) of the transferred graphene shows a single sharp band at

283.8 eV due to the sp2 C-C bonds of the graphene sheet and confirms that the

chemical state of the transferred graphene is unaltered compared to the original

one.

We also performed SPR measurements to help us determine the exact thickness of

the graphene layer, since previous studies showed a linear shift between the SPR

resonance and the number of graphene layers[147, 143]. The surface plasmon res-

onance curves measured at λ = 670 nm before and after graphene transfer can be

109


Figure 5.3: a) Raman spectra of the transferred graphene (blue) and the original

graphene on Ni (red); b) C1s high resolution XPS spectra of graphene before (solid

red) and after (dashed blue) transfer.

seen in Fig. 5.4. The refractive index of graphene in the visible range was estimated

to be n=3+iλC/3, where λ is the wavelength in µm and the constant C equals 5.446

µm−1 [148]. The imaginary part of the refractive index is k=1.216 at this wave-

length and was used to fit the experimentally determined SPR curves. From the

angle shift, a thickness of 1.02 nm can be deduced in both cases, which corresponds

to 3 graphene layers. The number of graphene layers can also be concluded from

the intensity of the I2D/IG (see Fig. 5.3 a) which gives an estimation of 3 graphene

layers, in agreement with the SPR results.

For the graphene on quartz samples the number of graphene layers and the

quality of the graphene sheet was provided by the supplier (ACS materials) and

confirmed by us using Raman spectroscopy. The measurements showed that we had

a single layer high quality graphene sheet. Fig. 5.5 compares the Raman spectra

provided by the supplier (see part a) and Raman spectra performed by us (see

part b). Both graphs exhibit a sharp 2D band with an intensity of I2D/IG which

corresponds to a monolayer graphene. Furthermore the low intensity of the D peak

indicates that the density of defects is very low.

Finally when depositing metallic NPs on the graphene interfaces, SEM images

should be regularly performed to monitor the size and shape of the NPs as well as

their adhesion to the surface. Graphene makes the EBL technique more challeng-

ing since metallic NPs tend to slightly shift during the lift off process from their

predefined position probably due to capillary forces [149]. This explains the weak

adhesion between graphene and the thin Cr layer used. Fig. 5.6 displays the SEM

images of the successful samples with metallic NPs deposited on graphene interfaces

which were later investigated through optical measurements.

110

5.5 Localized surface plasmon resonance of the graphene based coupled NP/film

interface

Figure 5.4: Surface plasmon resonance curves of Au surfaces before (black) and

after (blue) transfer of graphene: experimental data (dotted lines); theoretical SPR

curves (full lines) [nprism= 1.52, nT i = 2.36+i3.47 (d=5 nm); nAu = 0.197+i3.67

(d=50 nm); nAg = 0.14+i4.581 (d=38 nm); ngraphene = 3.0+i1.216; l=670 nm.

5.5 Localized surface plasmon resonance of the

graphene based coupled NP/film interface

We studied the plasmonic properties of two interfaces, one with Au NPs directly

on 50 nm Au film, and the second with Au NPs on graphene coated Au film as

shown in the schematic in Fig. 5.7. We performed extinction spectroscopy on the

two systems using the UV-VIS home made setup. Fig. 5.8 displays the extinction

spectra under normal incidence for the different gratings fabricated onto graphene

coated Au surface (full lines) and Au film (dotted lines). Each of the spectra observed

for both systems is characterized with two resonances, a low wavelength sharp mode

and a second band at higher wavelength, ❧2=770 nm. The second band is however

rather broad and not well defined for all the investigated interfaces. We notice that

the presence of graphene did not change the nature of the low wavelength resonance

as we can still observe a mode almost independent of the diameter. It has a sharp

full width at half maximum (FWHM) which decreases with the increase in the

diameter of the NPs. An interesting feature of this mode is that it is sharpened in

the presence of graphene and its FWHM decreases from 64 to 50 nm for NPs of 140

nm in diameter. For comparison, the FWHM for the band at ❧2 is about 250 nm

for this array. The presence of graphene induces two main differences:

1) a slight sharpening of the low wavelength mode which is appealing for sensing

applications,

2) a consistent blue shift of ∼ 13 nm in the resonant wavelength observed for the

spectra of NPs on graphene compared to NPs directly on Au film.

111


Figure 5.5: Raman spectroscopy a) provided by ACS materials and b) measured

in our lab on graphene on quartz substrates, showing a good quality monolayer

graphene.

Figure 5.6: SEM images of graphene-based SPR decorated with Au NPs by EBL

with center-to-center distnace of 300 nm. The particles are 50 nm in height and

80 nm (a), 110 nm (b), and 140 nm (c) in diameter.

The recorded blue shift was not expected since graphene exhibits a high index of

refraction which a priori from an optical point of view should result in a red shift of

the plasmonic modes.

To understand the physical reason behind this shift, our colleague Gaëtan

Lévêque performed numerical simulations using the Green’s tensor method on a

single Au NP, deposited either onto glass coated with 50 nm Au film or on glass

coated with 50 nm Au and post-coated with graphene (glass/Au/graphene) with

illumination in normal incidence to the substrate. The thickness of the graphene

layer was chosen as 1 nm since experimentally, we determined 3 monolayers. The

optical constants of Au were taken from Johnson and Christy [121]. The numeri-

cal extinction spectra for the different interfaces with Au NPs of 50 nm in height

and varying particle diameter (80, 110, and 140 nm) can be seen in Fig. 5.9 for the

systems with and without graphene spacer. Simulations showed that both systems

112

5.6 Interaction of graphene with metals and the tunability of the optical properties

Figure 5.7: Schematic illustration of the different interfaces investigated: a

graphene-coated Au thin film decorated with Au NPs array; b Au NPs array

directly deposited onto thin Au film without the graphene spacer layer.

with and without graphene exhibit a low wavelength mode with a slight dependence

on diameter. However numerical results suggest that graphene does not induce any

blue shift and instead a slight red shift is observed, which is theoretically expected

with graphene’s high index of refraction.

To verify that the nature of the low wavelength mode is not affected by the

presence of graphene and to further understand the experimental results, we plotted

the electric field maps at the resonant wavelength of 524 nm for a vertical (see

Fig. 5.10 a) and horizontal section (see Fig. 5.10 b) of a 110 nm NP placed on top

of the graphene coated Au film. The field maps show that the nature of the low

wavelength mode is not changed when graphene is added as a spacer layer, where

again similar to the case of Au NPs directly on Au film, two hot spots are observed on

the top of the NPs. This is the result of the interference between the field scattered

by the NP gratings and the reflection of this very same scattered field by the Au film.

The position of the hot spots in contact with air pushes the resonance strongly to the

blue compared with NPs directly on glass. As clearly seen, numerical simulations

did not provide any optical based explanation about the blue shifted plasmon band

observed experimentally, which proves that the origin of this blue shift cannot be

explained with electromagnetic theory alone. For this reason, we believe that the

blue shift induced by graphene comes from charge transfer between graphene and

the Au NPs, which would modify the LSP frequency through a modified free carrier

density and plasma frequency. Further investigation of this assumption will be

presented in the following sections.

5.6 Interaction of graphene with metals and the

tunability of the optical properties

In a theoretical study, Giovanneti et al in 2008 [150] used first-principles calcu-

lations to perform density functional theory (DFT) calculations on the adsorption

of graphene placed on different metal surfaces. They showed how the contact of

graphene with metals could significantly alter the electronic properties. They in-

vestigated a wide set of metals: Al, Co, Ni, Cu, Pd, Ag, Pt and Au, showing that

this set of metals could be divided in two classes. The first class consists of Co, Ni,

113


Figure 5.8: Extinction spectrameasured in air of the Au surface (dashed lines) and

graphene-modified Au surface (full lines) decorated with Au NPs of 50 nm (black),

80 nm (blue), 110 nm (green), and 140 nm (red) in diameter, 50 nm in height and

center-to-center distance of 300 nm. The signal was collected with a ×10 objective

with a numerical aperture of NA=0.15. The reference for calculating the extinction

is taking on the gold film outside the arrays. The optical extinction spectrum of

80 nm Au NPs directly fabricated on Au film could not be resolved.

and Pd and induces massive changes in the electronic structure of graphene through

chemisorption, where the graphene band structures are highly altered to an extent

they possess a mixed graphene-metal character. The second class consists of Al,

Cu, Ag, Au, Pt and is of greater interest fo us, since they exhibit weak adsorption

to graphene. However, even with such a weak bonding, the metal causes a shift in

the Fermi level away from the conical points. These shifts result in graphene doping

either with holes or electrons. Typically in a freestanding graphene, the Fermi level

coincides with the conical point. However as graphene is adsorbed to the second

class of metals the Fermi level is shifted depending on the type of doping as shown

in Fig. 5.11. When electrons are donated by the metal to graphene an upward shift

is induced in the Fermi level and the graphene becomes n-doped, while for the case

when holes are donated by the metal, the Fermi level downshifts and graphene is

p-doped. For a specific difference between the work function of graphene and the

metal, the doping can occur when the two are brought close enough together. The

sign (i.e the type of charge carriers) and amount (i.e the concentrartion of charge

carriers) of doping can be deduced from the difference of the metal and graphene

work functions [150]. Giovanneti et al developed a model to predict the directions

of the Fermi level shifts based on the work functions of graphene and clean metal

114


Figure 5.9: Computed extinction spectra of a single cylinder particle, diameter 80

nm (blue), 110 nm (green), and 140 nm (red), thickness 50 nm, placed on the Au

surface (solid lines), or on the graphene-modified Au surface (dashed lines).

surfaces. They also predicted how the metal work functions might be modified by

adsorption of graphene. When graphene is in contact with a metal, the work func-

tion of both the graphene and the metal is modified and an electron transfer between

the two occurs in order to equiliberate the Fermi levels. It is also important to note

that besides the electron transfer, there is also a chemical interaction between the

two which plays a role [150]. According to this model and to the work functions

of different metals, the contact with graphene can iduce n-doping of graphene for

Al, Ag, and Cu , while Au and Pt are expected to p-dope graphene. Furthermore

several studies on hybrid graphene metallic NP systems have been reported since

2012 [151, 152, 153, 154]. Evidence of changes in the charge carrier density and

shifts in the Dirac point by hot electron doping of graphene through excitation of

metallic nano-antennas was brought by Fang et al [151]. Also in a recent work,

Gilberston et al [152] were able to study the plasmon-induced hot carrier dynam-

ics in hybrid Au NP/graphene structure, using a femto second pump-probe setup.

This confirmed the charge transfer between graphene and metallic NPs. Graphene

exhibits a broad band light absorption making it very appealing for light material

interaction applications. However its nanometric thickness remains a challenging

limitation.

This was the main motivation behind the rise of hybrid systems with graphene

and optically active nanomaterials as quantum dots and metallic NPs [152]. Re-

search on active plasmonic devices based on electrostatic gating of hybrid semicon-

ductor/ metallic NPs systems is highly interesting, however it has been limited to

the Tetrahertz region [155, 156, 157]. Translating these devices to optical frequen-

cies is very challenging, since the response of semiconductors to electrostatic doping

is only efficient for the low frequency range [158]. With the discovery of graphene

as a zero-bandgap semiconductor [133, 159], this challenge might be resolved since

graphene can be doped over a broad frequency range [160, 161]. In 2012 Kim et al,

[153] showed the possibility to electrically control the plasmonic properties of a Au

115


Figure 5.10: a) Computed distribution of the electric field inside a vertical section

of the 110-nm diameter particle on the Au substrate, at the resonant wavelength

❧ =524 nm. Color scale electric-field time-averaged amplitude, normalized to the

incident plane wave amplitude; green vectors: electric field real part; cyan vectors:

electric field imaginary part. b) Computed distribution of the electric field in a

horizontal section 25 nm above the Au interface, same wavelength.

nanorod by electrostatic gating of a graphene sheet deposited on top of the nanorod.

The electrical gating of graphene showed the possibility to modulate the position

of the resonance, increase the quality factor by 30%, and increase the resonance

scattering intensity by 30% [153]. They explained these changes by variation in the

real (ε′g) and the imaginary part (ε′′

g) induced by electrical gating [153]. They also

showed that even few graphene electrons contribute to plasmon modulation when

present in the plasmon hot spots.

Even with all the work already done, several questions remain to be answered.

Would experimentally the contact with metal induce doping in graphene as theoreti-

cally predicted? Is the doping induced by metallic NPs sufficient enough to result in

the modulation of the plasmonic resonance compared to electrostatic doping? Is the

doping by metallic NPs responsible for the unexpected wavelength blue shifts and

sharpening in the plasmonic resonance which was observed in the graphene based

coupled NP/film system (see Fig. 5.8)? What would happen if graphene is n-doped,

as this was not previously before?

The first question was answered by an experimental study conducted by Lee et

al in 2011 [162]. They studied the interaction between metal film and graphene and

the dependence of this interaction on the number of graphene layers. They studied

induced doping of graphene by deposition of Ag or Au NPs formed by evaporation of

4 nm Ag or Au film on exfoliated graphene samples on a SiO2(300nm)/Si wafer. For

Ag NPs, the G band upshifts by 3.3 cm−1 while the 2D band downshifts by 4.5 cm−1

after the deposition of the NPs which indicates n-doping of graphene by the Ag NPs.

116


Figure 5.11: Schematic of the cone representing the electronic structure of

graphene, the cone center is the "Dirac point"; which is equivalent to the "Fermi

level" in graphene, and the effect of n and p-doping on this level.

In the case of Au NPs deposition, both the G-band and the 2D band upshift which

indicates a p-doping of graphene. The experimental evidence of graphene doping

by deposition of metal NPs through Raman spectroscopy came in agreement with

the theoretical assumptions, where Au NPs p-doped graphene and Ag NPs n-doped

graphene [162]. This motivated us to pursue our assumption that the blue shift

observed in LSPR in the presence of graphene could be the result of p-doping where

both the workfunction and the density of charge carrier in graphene and in the NPs

could be altered.

Such an active plasmonic system is very appealing and has high potential for

several applications especially sensing. To further validate this hypothesis, we de-

cided to investigate the plasmonic response for Ag in the presence of graphene and

compare it to that of Au and to perform Raman spectroscopy measurements to

evidence the doping. If our assumptions are true, Ag NPs should show a red shift

in LSP resonance when graphene is added. However coupled NPs/film interfaces

studied in this thesis can be a rather complicated system for such a study. So we

decided to decrease the degree of complexity and study 2 systems: one with Ag

NPs/monolayer graphene/quartz substrate and the other with Au NPs/monolayer

graphene/quartz substrate. We compare them with their reference systems i.e an

identical interface but without a graphene layer. We performed LSPR on both sys-

tems as well as Raman spectroscopy. The obtained results are highly interesting

and are presented in the following section.

5.6.1 p-doping of graphene and LSPR blue shift

We prepared a sample with Au NPs of diameters 140, 170, and 200 nm on top

of graphene monolayer/quartz substrate as shown in the schematic Fig. 5.12. The

200 nm array of NPs was damaged since depositing NPs by EBL on graphene can be

a challenging process as we discussed earlier. For this reason, we will limit our study

to the 140 nm and 170 nm NPs. We performed extinction spectroscopy as well as

Raman measurements to study the influence of a graphene layer on both plasmonic

properties and the graphene doping in such a system. In order to examine if Au

117


Figure 5.12: Schematic of the Au NPs based interface showing Au NPs of different

dimensions fabricated by EBL on top of mnolayer graphene/quartz substrate.

NPs induced p-doping in the graphene layer as theoretically expected, we performed

Raman measurements on the NPs and on regions with no NPs. Since the expected

shift in the Raman measurements is too small and gets to a maximum value of

5 cm−1, we had to perform around ten measurements both on the regions with

NPs and on regions without NPs to minimize the possibility of experimental errors.

Fig. 5.13 (a) shows that the G band upshifts from 1584 cm−1 to 1587 cm−1when the

NPs are added, and Fig. 5.13 (b) reveals also an upshift of around 4 cm−1 of the

2D band for zones with and without NPs. The upshifts of the G band and the 2D

band prove that as predicted theoretically, the contact of Au NPs with graphene

leads to a transfer of electrons from the graphene sheet to the metallic NPs and thus

p-doping of graphene occurs.

It is important to note that doping of graphene should theoretically lead to a

decrease in the I2D/IG ratio which was not observed in our experimental results.

The reason why no decrease in this ratio was observed is probably because of the

two different and seperate phenomena which are taking place at the same time: i)

first doping of graphene by contact with metal NPs and ii) second Au NPs lead to

Surface enhacement of the Raman signal (SERS). Metallic NPs have been used for

a long time in SERS [163]. In the case of graphene the G band and the 2D band

are of two different natures and thus their Raman enhancement by the presence of

the metallic NPs is not proportional which explains why the I2D/IG did not show

the expected decrease when graphene was doped.

As for the optical properties Fig. 5.13 (c) shows the extinction spectroscopy

measurements of Au NPs/ monolayer graphene (red spectra) compared to similar

NPs fabricated directly on quartz (black spectra). The unexpected blue shift ob-

served previously for coupled NP/film systems was again observed for this simplified

interface. A consistent blue shift of around 13 nm was observed for both the 140

nm and 170 nm NPs. Moreover the presence of graphene resulted in a slight sharp-

ening of 10 nm (fitted by a lorentzian) in the resonance peak. The wavelength blue

118


Figure 5.13: a) and b) Raman spectra showing the G peak and 2D peak respec-

tively for Au NPs/graphene quartz the black arrows are guidelines to show the

upshift of both G peak and the 2D peak when measured on regions with Au

NPs compared to thopse without Au NPs. b) Extinction spectroscopy: red is for

NPs/monolayer graphene/quartz, and black is for NPs/quartz, a blue shift ∼13

nm is recorded for 140 and 170 nm NPs.

shift and the sharpening of the plasmonic resonance are both in agreement with

the previous study of Kim et al, which demonstrated how electrostatic gating shifts

the Fermy energy and lead to modification in the optical transitions of graphene

[153]. Several studies investigated the dependency of graphene’s dielectric constant

on the doping level of graphene [164, 165]. An increased p-doping induces changes

in both the imaginary and the real part of the dielectric constant of graphene. This

changes the effective index of refraction of the surrounding media, and thus leads to

shifts in the plasmonic resonance. As the hole density increases, the Fermi energy

decreases and the imaginary part of the dielectric constant ε′′g is also reduced as a

result of blocked interband transitions. Since ε′′g represents the absorption or the

losses in the medium, its decrease explains well the sharpening in the plasmonic

resonance. Increasing the quality factor of plasmonic resonances through graphene

doping can be very appealing for high sensitivity applications. These results seem

119


to be very promising when linking the p-doping of graphene measured by Raman

to the unexpected blue shift and sharpening of graphene measured through UV-VIS

spectroscopy. Especially that doping induced by metallic NPs is not expected to be

as efficient as electrostatic doping. However a conclusive relation between graphene

doping by metal contact and plasmonic modulation of metallic NPs cannot be con-

firmed. To further validate this hypothesis we performed an identical study on Ag

NPs. Since Ag NPs induce n-doping of graphene, the optical response of such an

interface can be very informative.

5.6.2 n-doping of graphene and LSPR red shift

To further investigate the dependency of the LSPR shift on the doping of the

graphene sheet, we investigated a system with Ag NPs deposited on a graphene

monolayer/quartz substrate as shown schematically in Fig. 5.14. Theoretically and

Figure 5.14: Schematic of the Au NPs based interface showing Au NPs of different

dimensions fabricated by EBL on top of mnolayer graphene/quartz substrate.

based on the work function of Ag compared to that of graphene, Ag NPs are expected

to n-dope a graphene sheet when the two are brought in contact [150], as electrons

transfer from the NPs to graphene. This makes Ag a suitable candidate to study

the influence of the different types of graphene doping on the plasmonic properties

of metallic NPs. Similarly to the case of Au NPs, we also performed both Raman

and extinction spectroscopy. Fig. 5.15 (a) and (b) indicate that the deposition of

Ag NPs leads to n-doping of the graphene sheet, where the G band upshifts by ∼3.8 cm−1 and the 2D band downshifts by ∼4 cm−1. Another important note which

can be concluded from the Raman spectra is the induced mechanical defects in the

graphene sheet after the deposition of the NPs. The level of defects can be probed

using the D band intensity and it can be seen that after the deposition of the NPs,

the almost free graphene sheet exhibits an increased intensity of the D band and

thus an increased density of defects. The defect in graphene can arise from both the

highly energetic electrons impinging on the sample and the deposition of metallic

120


Figure 5.15: a) and b) Raman spectra showing the G peak and 2D peak respec-

tively for Au NPs/graphene quartz the black arrows are guidelines to show the

upshift of the G peak and the downshift of the 2D peak when measured on regions

with Au NPs compared to those without Au NPs. c) Extinction spectroscopy: red

is for NPs/monolayer graphene/quartz, and black is for NPs/quartz, a blue shift

∼13 nm is recorded for 140, 170 and 200 nm NPs.

films using PVD during the EBL process. However even in the presence of defects,

we can still study and monitor the doping of graphene without any problem.

As for the extinction spectroscopy, Fig. 5.15 (c) presents that the spectra for

the NPs deposited on top of graphene (red lines) show a red shift and a broadening

compared to those fabricated directly on quartz (black lines). From an optical point

of view based on graphene’s high index of refraction the plasmonic modes of the

Ag NPs are expected to red shift. The experimental results are in agreement with

those predictions where graphene induced a consistent and rather large∼29 nm red

shift for different NPs compared to their analogue on glass substrates. The red shift

observed for Ag NPs is much higher than the blue shift observed for Au NPs. This

leads to the conclusion that the modulation of the plasmonic shift resulted from

both the change in the optical constant and the charge transfer of electrons from Ag

121


to graphene. These results validate the assumption made earlier on the link between

the direction of the LSPR shift and the type of graphene doping.

Besides the red shift, the presence of graphene broadens the peaks of the dif-

ferent NPs, and the FWHM increases in the presence of graphene from 24 nm up

to 82 nm. The broadening depends on the size of the NPs where the larger the

NPs are the more significant the broadening is. The broadening results from the

high absorbtion of a graphene sheet. The absorption increases even more as the

imaginary part of the dielectric ε′′g increases due to n-doping of graphene induced by

Ag NPs. The increase in the broadening versus the increase in the size of the NP

is the result of increased variation in ε′′g , since larger NPs (i.e higher contact area)

induce a higher doping level of graphene.

The comparison of the influence of graphene on the plasmonic properties of Au

NPs and Ag NPs is highly interesting. It reveals that graphene can be doped by

contact with metallic NPs, and even small variation in the charge carrier density, as

those induced by metallic NPs can modulate the plasmonic resonance. A consistent

behaviour was observed for the two metals: p-doping of graphene resulted in a blue

shift and sharpening in the LSP resonance of Au NPs, while n-doping resulted in a

big red shift and broadening of the LSP resonance. Several plasmonic parameters

can be controlled through doping of graphene in the vicinity of metallic NPs: mainly

the resonance wavelength and the quality factor. These results further validate the

importance of hybrid metallic NP/graphene systems. Previous studies on the mod-

ulation of plasmonic response by graphene doping was limited to the near infra red

region [153]. However in this study, we showed that this behavior can be translated

to optical frequencies determined by the geometry and size of the NP, since graphene

doping is efficient over a broad frequency range. The possibility to modulate the

quality factor is very appealing for sensing measurements and will be addressed in

the next chapter. The possibility to modulate the resonance wavelength, even for

low level of graphene doping paves the way towards graphene based optoelectronic

devices in the optical frequency range. The work performed here on plasmonic mod-

ulation by graphene doping is thus only the tip of the iceberg in this research topic,

and a lot is still left to be done. Electrostatically doping graphene while performing

extinction measurements with an exact quantitative representation of the doping

level and a detailed investigation on the best configurations to couple wih graphene

should be performed. These topics will be pursued and investigated in the near

future.

5.7 Conclusion

In this chapter we discussed the possibility to use graphene, a newly discovered

2D material with extra ordinary optical and electronic properties as a subnanometer

spacer layer. Implementing graphene in coupled NP/film interfaces shows promis-

ing results both for sensing enhancement and for higher modulation and tunability

122

5.7 Conclusion

of these systems. The unexpected resonance shifts we observed encouraged us to

investigate graphene doping through contact with metallic NPs. Our experiments

showed that Au NPs can induce p-doping in graphene and a blue shift in the LSPR

resonance of the NPs, while Ag NPs can induce n-doping of graphene and red shift

of the LSPR. The dependence of the LSPR resonance on the doping level of the

graphene layer underneath could be very promising, especially with the possibil-

ity to electrostatically dope graphene. These results are very appealing towards

graphene based optoelectronic devices.

123

6 Coupled NP/film systems forSERS and RI sensing

6.1 Introduction

Throughout this manuscript, we have thoroughly discussed the optical prop-

erties of coupled NP/film systems. We showed how MIM structures can be rich

plasmonic interfaces with an interplay between localized and delocalized plasmonic

modes. We also showed the possibility to use unconventional novel material such as

graphene as an ultra thin spacer layer, and we investigated the different plasmonic

modes and the effect of graphene on these systems. After an extensive investigation

of coupled NP/film systems, it was natural to study the potential of these systems in

typical optical applications, as refractive index (RI) sensing and Surface Enhanced

Raman Spectroscopy (SERS). In this chapter, we compare RI sensing measurements

for different interfaces with and without a gold film, and with different spacer layers.

We show that coupled NP/film systems may enhance the quantitative performance

of the sensors by both increasing the sensitivity and sharpening of the plasmonic

mode. On the other hand, graphene is known to have a distinct Raman fingerprint,

which makes it a suitable candidate for studying the enhancement of the Raman

signal. This encouraged us to perform SERS measurements of Au NPs on graphene

deposited on different substrates. The enhancement of the electric field induced

by the coupling of the NPs and the film resulted in higher enhancement factors of

the graphene signal for these interfaces compared to different substrates without a

metallic film.

6.2 Refractive index sensing

6.2.1 Physical background

Since the rise of plasmonics these systems have been used in multidisciplinary

applications [39], with one of the earliest and most common application being sens-

ing. Extensive research on optical sensors based on SPR and LSPR plasmonic

modes for the detection of trace molecules and molecular interactions in both bio-

logical and chemical systems have been conducted. In a recent study we compared

the performance of both SPR (on au film) and LSPR (for Au NPs) resonances as

125

Chapter 6 Coupled NP/film systems for SERS and RI sensing

optical sensors and we investigated the advantages of each [166]. The electric field

enhancement and the ultra sensitivity of SPR and LSPR resonances to the sur-

rounding medium is the reason behind the increased interest in plasmonic based

biosensors. The last decade has witnessed fast advancement in nano-fabrication

techniques which allowed the fabrication of all kinds and sizes of nanostructures.

This resulted in a remarkable advancement in LSPR based biosensors. One of the

challenges for comparing the performance of different LSPR sensors comes from the

inconsistencies in the nomenclature found in the numerous articles on this topic [79].

For this reason, we will discuss the physical background of LSPR sensors and the

method by which the quality of our sensors is quantified.

The main motivation behind LSPR biosensors comes from the dependency of

the latter on the surrounding medium. To better understand the relation between

the surrounding index of refraction and the wavelength shifts, a functional form of

LSPR resonance dependence on the dielectric medium can be derived [167]. Starting

with the simplified dielectric function equation:

ε = 1 − w2p

w2(6.1)

where ε is the dielectric constant, and wp the plasma frequency, and using the

resonance condition ε = −2εm, where εm is the dielectric constant of the surrounding

media, the equation becomes:

wmax =wp√

2εm + 1(6.2)

where wmax is the resonant frequency of the LSPR mode. This relation can be

rewritten in terms of wavelength and index of refraction, knowing that λ = 2πc/wand εm = n2 eq 6.2 becomes:

λmax = λp

√2n2

m + 1 (6.3)

where λmax corresponds to the resonant wavelength, and λp to the wavelength of

the bulk metal plasma frequency. Eq 6.3 shows the dependence of the resonant

wavelength on the index of refraction where an approximately linear relation exists

between the two for optical frequencies.

Refractive index sensing is one of the simplest yet very informative applications

on the quality of LSPR resonances. It is based on detecting the changes in the

bulk index of refraction surrounding the NPs through wavelength peak shifts of the

extinction spectra. With the linear relation between the wavelength shift and the

index of refraction, the refractive index sensitivity S can be easily introduced as:

S =dλp

dn(6.4)

126


and is represented in nanometers of peak shift per refractive index unit (nm/RIU).

Since the performance of the LSPR sensor depends not only on the sensitivity but

also on the line width of the resonance, another quantity is introduced to better

represent the performance of sensors and is called the Figure of Merit (FoM). FoM

is defined as the sensitivity divided by the full width at half maximum (FWHM) of

the resonance [168] and is commonly used to characterize the sensing capabilities of

a system. High values of FoM is thus an indicator for good sensor performance and

good readability. The sensing potential of our different substrates will be compared

using both the sensitivity S and the FoM.

To perform our sensitivity measurements, we changed the index of refraction

surrounding the NPs by using mixtures of water and glycerol at different concentra-

tions, giving indices of refraction varying from n=1.33 to n=1.47. All the measure-

ments were performed under normal illumination on the Nikon spectrometer (see

Chapter 2). To better understand the advantage of the coupled NP/film systems and

the importance of introducing a spacer layer, a comparison with reference systems

was necessary. For this reason, we also performed similar sensitivity measurements

on an interface with Au NPs arrays exhibiting various diameters and 300 nm peri-

odicity deposited directly 1) on a quartz substrate, and 2) on a 50 nm Au film (used

for the experiments in the previous chapters). Au NPs on a quartz substrate are

characterized by a parallel (longitudinal) LSP mode. This mode shows high sensi-

tivity to the change in the index of refraction, however it is a rather broad band

(FWHM ∼ 200 nm) or even more. As the size of the NPs increases, the sensitivity

also increases, however this is associated with a broadening of the peak, and thus it

is not possible to increase their FoM. The highest FoM which was recorded for this

interface was for the 140 nm NPs and assessed to 1.9.

For the second reference interface with the NPs deposited directly on the Au

film without any spacer layer, it is charecterized by a low sharp mode around 520

nm with the hot spots at the top of the NPs in contact with air. As predicted

by Hohenau and Krenn, the LSPR peak sharpness should lead to highly sensitive

sensors [30]. For both its sharpness with a FWMH = 64 nm and almost indepent

of the NPs size, and its localization at the interface to the surrounding medium,

the possibility to use it for sensing applications was investigated. Even with the

remarkably decreased FWHM of this mode compared to the parallel dipolar mode

for NPs on glass, the FoM cannot be increased since the measured sensitivity is much

lower. The highest FoM was recorded for the 140 nm NPs and with a sensitivity S =

133 nm and a FoM = 2.1[169]. Adding an Au film resulted in only a minor increase

in the FoM compared to glass substrate, however in the following subsections we

will discuss in details sensing measurements and show how the FoM can be further

increased and even doubled when a spacer layer is added.

127


6.2.2 Au NPs on graphene coated Au film/glass substrate

We showed in the previous chapter that Au NPs on a graphene sheet induce

p-doing of graphene and 14 nm sharpening in the plasmonic mode. We performed

sensing measurements on this graphene based sensor Fig. 6.1 (a) reveals that the

position of λ1 is shifting to higher wavelengths with the increasing refractive index.

The change in the position of λ1 and ∆λ1, exhibits a linear dependency as a function

of the refractive index of the surrounding medium. The sensitivity is determined

Figure 6.1: a) Extinction spectra of the graphene-modified SPR surface decorated

with Au NPs array of 140 nm in diameter for different refractive indexes n of glyc-

erol/water mixtures: 1.00 (black), 1.33 (gray), 1.37 (blue), 1.40 (magenta), 1.44

(red), and 1.47 (green). b) Shift of the low wavelength LSPR peak dependingon

the refractive index of the surrounding medium.

from the slope (see Fig. 6.1 (b)) and increases as the thickness/diameter ratio of

the plasmonic interface decreases. Graphene-modified Au films coated with Au

NPs of 140 nm in diameter exhibit a sensitivity of 139 nm/ RIU, whereas the same

systems coated with Au NPs of 110 and 80 nm display sensitivities of 66 and 34 nm/

RIU, respectively. The observed sensitivity of 139 nm/RIU is comparable to other

plasmonic structures with a resonance band between 500–600 nm [170, 171, 172]. For

the interface with Au NPs of 140 nm in diameter with graphene spacer, the FWHM

is equal to 50 nm and results in a FoM of 2.8 when fitted by a Lorentzian function

as compared to only 2.1 without spacer. The presence of graphene thus results in

a 33% enhancement of the FoM compared to the system with no spacer, and 47%

increase compared to NPs directly on glass. Besides the p-doping of graphene due

to its contact with Au NPs results in the sharpening of the plasmonic mode and a

small increase in the sensitivity and thus enhancement of the FoM. This can be very

promising for future applications on graphene based optical sensors, especially with

the possibility to electrostatically dope graphene.

128


6.2.3 Au NPs on MIM interface

Fig. 6.2 (a) presents the evolution of the extinction spectra versus the increase

in the index of refraction of the surrounding medium of an array NPs with 200 nm

diameter and 300 nm periodicity deposited on a coupled NP/film system with a 6

nm SiO2 spacer layer (investigated in chapter 3). This experiment illustrates how

the increase in the index of refraction induces a red shift of the LSP mode and a

blue shift of the glass-side SPP mode, leading to a coupling of these modes and a

sharp resonance around 550 nm, linearly dependent on the change in the index of

refraction. However the gap mode is almost insensitive to changes in the index of

refraction which is to be expected since the energy is stored in the gap.

Sensitivity was calculated for NPs with different diameters and a 300 nm grating

periodicity (see Fig. 6.2 (b)). Sensitivities of 186, 192 and 185 nm/RIU were recorded

for the 140, 170 and 200 nm NPs respectively. These sensitivities are remarkably

higher compared to similar NPs/film systems without a spacer layer, and to the

system with graphene spacer. On the other hand, this peak profile is significantly

sharpened compared to the case were no metal film is present and its full width

at half maximum (FWHM) is equal to 50 nm for the large NPs and even less for

the 140 nm NPs. Thus the two advantages of the metal-insulator–metal systems

in increasing the sensitivity and in sharpening the mode spectral width are very

promising for enhancing the plasmonic sensors. The FoM for the three different

diameters of NPs was calculated. The 140 nm NPs recorded the highest FoM of 4.2

which is more than double of the previous FoMs that we reported, thus more than

a 100% increase, while the 200 nm NP also recorded a high FoM equal to 3.78.

This high FoM in the presence of a dielectric spacer can be understood first

by the highly increased sensitivity, almost comparable with that of NPs directly

on glass, and second by the overlap of the LSP and the SPP modes resulting in a

significant sharpening of the peak. Numerical simulations have been performed for

the particle with 200 nm diameter, showing a faster red-shift of the low-wavelength

localized plasmon with the index of refraction than in the experiment. Fig. 6.2 (c)

put clearly into evidence, together with the electric field distribution for n=1.4 at

the low wavelength resonance (presented in the inset), that this is the same mode

with the two hot-spots at the top edges of the nanocylinder which shifts to the red

while the gap modes are more weakly affected by the change in refractive index, as

the light is concentrated inside the gap.

Fig. 6.2 (d) shows the evolution of the position of the low-wavelength mode

which is remarkably linear with n2. The refractive index sensitivity is 246 nm/RIU

for that mode arounf 520 nm, while the gap mode at 740 nm has a lower sensitivity

of about 38 nm/RIU. The discrepancy between the numerical and the experimental

value can be explained by the fact that the edges are sharper in the model than in

the experiment, where the shape of the particle is probably slightly conical.

Such a FoM value is higher than those reported for silver triangles (❧peak=564

nm; FoM=1.8)[170], silver cubes (❧peak=510 nm; FoM=1.6)[171], silver spheres

129


(❧peak=520 nm; FoM=2.2) [172], or gold spheres ❧peak=530 nm; FoM=1.5)[173].

This ranks our sensor among the highly sensitive, if not one of the highest LSP

sensors with plasmon band in the visible at 500 nm.[48] Indeed, most of the interfaces

with high FoM (4–16.5)[174] take advantage of the fact that higher sensitivities are

achieved with plasmon bands in the near-infrared of the spectrum (850–1,200 nm)

and of phase measurements.

6.3 Surface Enhanced Raman Spectroscopy (SERS)

Enhancement of the Raman spectroscopy with an increase in the Raman cross

section and thus in the Raman intensity was first discovered in 1977 [175, 176]. Two

decades later, enhancement factors of 1014 of the scattering cross section allowed

single molecule detection [177]. These enhancements are achieved when the local

electric field is increased, thus “hot spots” appear in the vicinity of metal NPs

arising from localized surface plasmon resonance [177, 66]. The detection limit was

the main reason of the increased interest in SERS; making it highly appealing for

sensing applications. Up till today no complete theory or model is available to

describe the SERS effect. However the scattering cross section enhancement is well

known to originate from chemical and electric field enhancement processes. The

chemical enhancement factor is due to a change in the environment of the molecules

but it does not exceed 100. For this reason, research on increased enhancement due

to electric field with different plasmonic particles is continuously emerging. The

electromagnetic theory of SERS will be briefly discussed in the next section.

6.3.1 The electromagnetic theory of SERS

The enhancement of the Raman intensity occurs by placing Raman-active molecules

within the near-field of the “hot spots” on metallic nanostructure. As previously

discussed in the first chapter small nanoparticles can exhibit a dipolar plasmon res-

onance when illuminated. The increase in the scattering cross section of the induced

dipole increases the intensity of the local field in the vicinity of the NP which leads

to an enhancement of the Raman signal. A complete description of the electromag-

netic theory of SERS can be found in [178], while for the scope of this thesis only a

brief overview will be discussed. For the Stokes process the power of the scattered

beam is

PS(υS) = NσRSI(υL) (6.5)

where N is the number of Stokes-active scatterers within the excitation spot, σRS is

the scattering cross section, νLthe excitation frequency, νSthe scattering frequency,

and I(υL) the intensity of the excitation beam. When the Raman active molecules

are placed in the near field of metallic nanoparticles the intensity PS increases which

130


Figure 6.2: a) Extinction spectra of the Au NPs with 200 nm in diameter and 300

nm in periodicity, for different refractive indexes n of glycerol/water mixtures:

1.00 (black), 1.330 (red), 1.3697 (light blue), 1.4016 (magenta), 1.4410 (green),

and 1.4616 (dark blue) b) Shift of the low wavelength LSPR peak as a function

of the refractive index of the surrounding medium; c) Numerical simulation of the

extinction spectra of the Au NPs with 200 nm in diameter and 300 nm in peri-

odicity, for different dielectric constants n2 of the NPs’ surrounding medium ; the

inset shows the distribution of the amplitude of the electric field in the polarization

plane for n=1.4 and ❧=610 nm ; d) Numerical shift of the low wavelength local-

ized plasmon with n2 and corresponding linear regression. The refractive index

sensitivity for the low wavelength mode is equal to 246 nm/RIU.

131


has two effects. First the Raman cross section σRS increases due to the changes in

the environment and becomes σSERS which constitutes the chemical enhancement.

Second and more importantly, PS is enhanced because of the increased electromag-

netic field due to excitations of localized surface plasmons and a crowding of the

electric field lines (lightning rod effect) at the metal interface[178]. Both the incom-

ing and emitted light fields are enhanced and the new enhanced total power due to

SERS becomes:

PS(υS) = NσSERSL(υL)2L(υS)2I(υL) (6.6)

where L(υ) is the electromagnetic enhancement factor, L(υ) =|Eloc(υ)|

|E0|with | Eloc |

being the local field amplitude at the Raman active site and | E0 |the incident electric

field. L(υL) ≈ L(υS) since the difference between the frequencies of the incoming

and scattaered photons is negligible. As a conclusion, the total SERS enhancement

scales to the fourth power of the field enhancement factor. A common expression

for the SERS enhancement R[178] is

R =| Eloc |4| E0 |4 (6.7)

Enhancing the electric field highly depends on matching the resonance between

the incident light and the resonance of the dipolar mode in metallic NPs. So the

SERS response to metallic nanostructures of different shapes and of different materi-

als can be very different. For this reason, extensive research on plasmonic materials

which would further increase the SERS enhancement is constantly ongoing.

6.3.2 Surface Enhanced Raman Spectroscopy of graphene

The ongoing research on Surface enhanced Raman spectroscopy allowed for

remarkable advancement of this technique where even single molecule detection is

now possible [177]. However transforming SERS into a commercial tool remains

very challenging especially with the limitation of its quantitative reproducibility.

On the other hand graphene a newly discovered material with its 2D structure

and its distinct Raman fingerprint seems to be a rising star in SERS studies. As we

discussed in Chapter 6, Raman spectroscopy is a powerful graphene characterization

tool. It can be used to identify the electronic structure, the number of layers, defects,

and the type of doping [179, 180, 178, 162, 133, 149]. Contrary to common SERS

substrates, there is no need for added molecules to detect Raman singal as the

enhanced Raman spectra of graphene itself can be studied. The main advantage in

using graphene as a bed test for studying the electromagnetic SERS enhancement

is that there is no need for normalization of different molecule concentration. This

makes the quantitative analysis easier and shows the possibility to use graphene as

a SERS probe [140].

132


When studying graphene SERS, several parameters should be considered. The

substrate on which graphene is deposited is highly important. Commonly most

reported works on graphene SERS use a silicon substrate with around 300 nm

oxide layer. The oxide layer allows both the visibility of single layer graphene

[181, 182, 183] and the enhancement in the Raman signal through thin film interfer-

ence effect [184, 185, 161]. In their study Lee et al [186], reported a 1.6 enhancement

factor due to the thin film interference effect. Nevertheless most experimental stud-

ies do not take this enhancement into consideration. The choice of the substrate is

important however it is not sufficient for any significant SERS study [186]. The need

for higher graphene Raman enhancement factors encouraged further research such

as tip enhanced Raman spectroscopy (TERS) [187, 188] where high enhancement

factors up to 6 x 103 were recorded. Also several studies on graphene SERS with

electromagnetic enhancement using nanostructures and their LSP resonances were

performed. In their study, Lee et al deposited Au islands on top of the graphene

sheets by evaporating thin Au films and the enhancement factor went up to a max-

imum of 120 for the G-band [186]. Actually a strong Raman signal can help in

visualizing the fine structure of graphene [189], and characterizing the defects by

enhancements of the D peaks [186, 190].

In one of the leading studies on resonant SERS of graphene, Schedin et al [149]

deposited Au nanodisks prepared by EBL on a graphene layer on top of Si/SiO2

substrate. They prepared NPs of two diameters: 140 and 210 nm, with a center

to center distance equal to 320 nm. They showed that the thin film interference

interaction is around 2.5 and more importantly that the deposition of Au NPs can

increase the enhancement factor up to 35 depending on the size of the NPs due to

the LSP resonance. They believed that the enhancement is mostly due to near-field

plasmonic effects in the vicinity of metal particles. They also performed numerical

simulations by deriving analytic expressions which well explain the physics of SERS

in 2D, which confirmed their experimentally reported enhancement factor. Following

this work, several studies on graphene SERS using ordered arrays of NPs were

performed. Nanostructures with different geometries, sizes and even different metals

with graphene deposited beneath or coating the nanostructures were investigated,

but the enhancement factor of the Raman peaks showed no remarkable increase

[191, 192, 186, 193, 194, 195].

6.3.3 Experimental results

We performed SERS measurements on different graphene based interfaces using

the LABRAM setup in our laboratory equipped with a 632.8 nm wavelength laser.

Since most studies on graphene SERS used Si/SiO2 as a substrate, we found it

interesting to investigate our coupled NP/film system with graphene as a spacer

layer. New samples were prepared with monolayer, bilayer, and trilayer graphene

deposited between the Au NPs and the Au film. Au NPs with dimensions similar to

those used throughout this thesis were prepared by EBL. Using a metallic film as a

133


substrate might lead to a higher enhancement of the Raman signal, even though the

thin film interference enhancement no longer exists, but we believe that the coupling

of NPs/film should lead to a higher enhancement of the EM field and thus higher

enhancement of the Raman signal. To provide a well controlled comparison with

the enhancement induced by similar NPs on different substrates we also prepared

identical Au NPs on top of monolayer graphene on a 1) glass substrate and 2) SiO2

(300 nm)/Si substrate. The enhancement factor (EF) of the the G-band and 2D-

band of graphene was measured by dividing the intensity of the peaks measured for

regions with NPs by the none of the reference i.e: regions with only graphene, as

shown in the following equation:

EF =INP s

Iref

(6.8)

For the NPs deposited on glass, the highest EF induced by the presence of the

NPs was recorded for the 140 nm NPs. It showed a maximum value of 17 for the

G-band and 25 for the 2D-band. While for the Si/SiO2 substrate, the value of the

EF was equal to 19 for the G-band and 28 for the 2D-band

When performing SERS measurements on graphene deposited on Au film,

the fluorescence effect due to Au is significant which makes it hard to measure the

intensity of the graphene bands. For this reason we used multi windows to measure

the 2D-band and the G-band and we performed a baseline correction on the different

spectra to better calculate the intensities of the Raman bands.

Monolayer graphene/Au film Fig. 6.3(a) and (b) shows the SERS measurements

of the G-band and 2D-band respectively for NPs of different diameters deposited on

monolayer graphene on a 50 nm Au film. The spectra show a remarkable enhance-

ment in the Raman bands due to LSP resonance in the presence of the NPs. The

EF varies depending on the size of the NPs Tab. 6.1 shows the calculated values of

the EF for different NPs for both the G-band and the 2D-band.

The calculated EFs reveal that the 140 nm NPs present a prominent enhancement

NPs size (nm) 110 140 170 200

G-band 14 433 26 35

2D-band 3 25 3 9

Table 6.1: Calculated enhancement factors of G-band and 2D-band for NPs of

different diameters deposited on monolayer graphene on an Au film.

in the Raman signal compared to the other NPs. It is difficult to explain why the

140 nm NPs show an enhancement higher than the other diameters, especially that

134


Figure 6.3: Raman measurements on graphene monolayer deposited on top of a 50

nm Au film, showing the enhancement for the regions with NPs compared to those

without NPs. The NPs are prepared by EBL and are of 80 to 200 nm diameter

with a periodicity equal to 300 nm. a) represents the G-band and b) the 2D-band.

they all have a similar extinction spectra as we discussed in the previous chapter (see

Fig. 3.8), with a single LSP mode around 520 nm. The 433 EF recorded for the G

band is very high compared to the EF reported by other groups, and to the reference

samples with NPs of similar dimensions prepared by EBL on top of graphene on

Si/SiO2, and on glass substrates [186, 162]. However in a study on graphene SERS

Khorasaninejad et al [196] reported a very high enhancement factor of the G-band

(∼ 890) for graphene deposited on top of a plasmonic structure with Ag rings and

crescents deposited on top af an Au film. This confirms that the coupling between

the metallic NPs and the metallic film results in the generation of hot spots and

the enhancement of the electric field, which contributes to a high enhancement of

the Raman signal. Eventhough we repoted an EF less than theirs, our plasmonic

interface is easier to design where no complicated nanostructures are required, and

the metallic NPs are made of Au where no oxidation problems should be addressed.

Another important remark is the difference in the EF between the G-band and

the 2D band, where for the 140 nm NPs for example, the G-band has a very high

enhancement of 443 while the 2D band shows an enhancement of only 25, the same

behavior is also observed for all other NPs. This can be explained by the difference

in the origin of the two Raman bands, and was also reported by other groups [196].

Bilayer graphene/Au film To study the effect of the number of graphene

layers on the SERS of graphene, we performed Raman measurements for Au NPs

wi similar dimensions deposited on bi-layer graphene on a 50 nm Au film. The EF

for the G-band and the 2D-band are summarized in Tab. 6.3, and several important

remarks can be concluded from these experimental results.

135


NPs size (nm) 110 140 170 200

G-band 31 40 35 25

2D-band 44 2.8 4

Table 6.2: Calculated enhancement factors of G-band and 2D-band for NPs of

different diameters deposited on binolayer graphene on an Au film

First the EF is much less than that recorded for monolayer graphene, the effect

Table 6.3: Calculated enhancement factors for the NPs of different diameters for

both the G-band and the 2D-band.

of the graphene thickness (or the number of layers) on the Raman enhancement has

been previously studied, and it was consistently shown that the enhancement de-

creases as the number of layers increase [162, 186]. The dependence of the graphene

Raman enhancement on the number of layers can be understood by the stronger

interaction between graphene and metals for single layer, this interaction decreases

as the number of layer increase due to the decrease of the Van der Waals forces

[197]. Second, the 140 nm NPs again showed the highest EF compared to NPs of

different dimensions. Eventhough we could not perfectly explain the dependency

of the enhancement on the diameter, we can confirm that the plasmonic resonance

of the 140 nm NP induces the highest enhancement, which was observed for both

monolayer and bilayer graphene. Finally, in contrary to monolayer graphene for the

bilayer we observe a D-band for the different NPs, this can be due to mechanical de-

fects during the NP fabrication process. However the enhancement of these defects

induced by the presence of NPs validates the possibility to use graphene SERS as a

graphene charecterization technique.

Finally for trilayer graphene, the EF was insignificant which confirms the im-

portance of the number of graphene layers on the graphene SERS phenomena. These

results confirm that coupled NP/film systems can be very advantageous in graphene

136

6.4 Conclusion

SERS, where the EF in the presence of the metallic film was one order of magni-

tude higher compared to other substrates. Eventhough the advantages introduced

by coupled NP/film was investigated only for graphene, we believe these interfaces

can be translated to other SERS measurents. The high EF factors recorded can

be considered of the highest for graphene SERS which can have very interesting

applications. It would also be interesting to perform SERS measurements on these

different interfaces with a green laser, especially that the plasmonic mode of cou-

pled NP/film systems have a LSP mode at 520 nm. A full theoretical study on the

origin of the high enhancement recorded for graphene in these systems could be a

key towards better understanding of the origin of the SERS phenomena and the

properties of graphene. Unfortunately this is beyond the scope of this thesis.

6.4 Conclusion

In this chapter we have shown the high potential of coupled NP/film in optical

applications. We showed that MIM interfaces with thin dielectric spacer are very

good sensors with FoM higher than most other RI sensors based on LSP resonances.

We also showed that the tunability of graphene induced by doping can also be very

interesting for future graphene based sensors. In the second part of this chapter,

we have investigated graphene SERS and we have shown that coupled NP/film

interfaces induced the highest enhancement in the Raman signal. A lot remains

to be done on graphene SERS where both measurements with lasers of different

wavelength and numerical simulations would highly increase our understanding of

the potential of these systems.

137

Conclusion

The aim of this thesis work was to invesitgate the optical properties of coupled

NP/film systems, and to establish a deep understanding of the underlying physics

behind these plasmonic interfaces and the different resonances associated with them.

We investigated MIM interfaces with thin SiO2 spacers and revealed the high poten-

tial of MIM structures. The MIM structure studied was considered a metasurface

due to the high tunability of the absorption, reflectance, and transmission in these

subwavelength interfaces. In a crossed comparison between the experimental and

numerical results, we investigated the nature of the different plasmonic modes. For

NPs fabricated directly on a metallic film, we showed the presence of a low wave-

length mode at 520 nm which is characterized by two hot spots pushed to the top

of the NPs. When a thin 6 nm SiO2 spacer is introduced a Fano resonance is also

observed besides the 520 nm mode, as well as a longitidinal LSP mode which is

observed only for small NPs. We studied the dependency on the diameter and peri-

odicity for these different systems, and we developed an analytical model to explain

the origin of the Fano resonance. After a full understanding of the optical prop-

erties of thin spacers, we gradually increased the spacer thickness and studied its

effect. This full systematic study can be considered as one of the most complete on

coupled NPs/film systems which enabled us to establish a thorough understanding

of the origin and dependencies of the different localized, delocalized, and Fano reo-

nances. A lot of work can still be done on MIM structures, to further optimize the

different parameters to serve application purposes. Tuning of the delocalized and

localized modes is possible by varying the different fabrication parameters (size of

NPs, periodicity, thickness of the spacer) and can result in the overlap and coupling

of different modes. Such a high tunability can result in exotic plamonic features

such as Fano resonances and can have a variety of highly appealing applications.

On the other hand the investigation which were performed on the graphene

based interface seems very promising, yet a lot of future work should be done. We

showed that experimentally the presence of graphene as a spacer layer can induce

an unexpected blue shift of the 520 nm mode, however this could not be reproduced

numerically. At early periods of this research we believed this blue shift was the

result of charge transfer between the graphene sheet and the Au NPs. At later

stages we investigated this hypothesis using both Raman and UV-VIS spectroscopy

and we showed that graphene can be either p-doped or n-doped by Au and Ag

NPs respectively. This doping results in a change in graphene’s dielectric constant

and thus explains the plasmonic shifts induced by graphene. With the possibility

to electrostatically dope graphene this can be a preliminary step towards graphene

139


optoelectronic devices.

Finally we studied the potential of coupled NPs/film systems in optical ap-

plications. We showed that MIM interfaces with thin dielectric spacer can be a

star candidate for optical sensors with a remarkably high FoM equal to 4.2 and

at a low wavelength. This high FoM is the result of the sharp mode resulting

from the coupling between the LSP and SPP modes, and the high sensitivity of

dielectric substrates. As for graphene based sensors the sensitivity is lower than

that of MIM structures, however the sharpening of the mode induced by p-doping

of graphene is highly appealing for sensing applications. Our measurements have

shown a good graphene based sensor with a FoM equal to 2.8, thus recording a 33%

increase compared to the NPs on metal film system. We also showed that coupled

NP/film systems result in a high increase in the G band enhancement factor of the

graphene Raman spectra, this is very important for graphene SERS measurements,

and graphene SERS probes. We believe further investigation and optimization of

the parameters of coupled NP/film systems can result in further advances in optical

applications, which remains the main objective of all research on plasmonics.

140

7 French Summary

Chapitre 1

Ce chapitre introduit les fondements théoriques de la plasmonique, c’est-à-dire

de l’interaction des métaux avec le champ électromagnétique aux fréquences op-

tiques. Les équations fondamentales gouvernant le comportement des plasmons

polaritons de surface (SPP) supportés par des films métalliques et celui des réso-

nances de plasmons de surface localisés (LSPR) de nanoparticules métalliques y

sont entre autres exposés ainsi qu’une revue des nombreux modes plasmoniques ob-

servés depuis l’émergence de ce domaine. Une attention particulière est portée à

la discussion des interfaces nanoparticules (NPs) métalliques / film métallique en

présence ou non d’une couche séparatice. L’état de l’art de ces systèmes hybrides

est détaillé avec une rapide mise en relief des étapes significatives qui ont jalonné le

développement de ce domaine ainsi que les principales applications envisagées.

Plasmons polaritons de surface:

Dans le cas d’interfaces diélectrique/conducteur, l’excitation électromagnétique

de modes plasmons délocalisés, appelés plasmons polaritons de surface (SPPs), peut

être induite. Ces ondes propagatives de surface sont excitées via le couplage du

champ électromagnétique incident avec les oscillations du plasma d’électrons dans le

conducteur. Ce couplage n’est possible que si le vecteur d’onde de la lumière incident

correspond à celui du SPP avec la même fréquence angulaire ✇. L’excitation directe

des SPP n’est donc pas possible s’il n’y a pas d’accord de phase. Les mécanismes

principaux pour lever le non-accord de phase et exciter le SPP reposent soit sur le

couplage à travers un prisme d’indice de réfraction plus élevé, soit sur le couplage à

travers un réseau de diffraction.

Résonances de plasmons de surface localisés (LSPR):

Une seconde forme d’excitations plasmoniques est connue sous la dénomination

de plasmons localisés de surface (LSP). Contrairement aux modes SPP, les modes

LSP sont non propagatifs et correspondent aux oscillations collectives des électrons

libres des NPs métalliques. Ils ont pour origine la diffusion de la lumière par des

141

Chapter 7 French Summary

NPs métalliques de taille sub-longueur d’onde et placées dans un champ électro-

magnétique oscillant. Les particules exercent en effet une force de rappel sur les

électrons libres ce qui conduit à un mode résonant amplifiant le champ électrique à

la fois à l’intérieur et dans le voisinage proche de la particule. Les résonances LSP

sont transverses et peuvent ainsi être excitées directement via une onde électromag-

nétique. L’argent et l’or présentent tous deux une résonance LSP dans le visible et

sont pour cette raison les matériaux plasmoniques les plus connus.

Présentation détaillée des différents modes plasmoniques

Avec les progrès en nanofabrication et en instrumentation optique, un large

éventail de résonances et de modes a pu être identifié dans diverses interfaces plas-

moniques. Nous discuterons brièvement de ces principaux modes plasmoniques et

plus particulièrement de ceux relevant de notre travail.

Modes localisés dans des nanosphères: Afin de comprendre les propriétés op-

tiques de nanosphères métalliques et les résonances plasmoniques associées, le re-

cours à la théorie de Mie permet de calculer les spectres d’extinction (absorption +

diffusion) de nanoparticules sphériques en fonction de leur diamètre.

Modes localisés dans des nanobâtonnets: Les nanoparticules anisotropes telles

que des bâtonnets ou des cylindres présentent deux modes à des longueurs d’onde

différentes : un mode transverse et un mode longitudinal. Le mode transverse est

excité via une polarisation le long du petit axe tandis que le mode longitudinal est

excité via une polarisation le long du grand axe.

Modes plasmons délocalisés symétriques et asymétriques dans des structuresde type IMI (Isolant/Métal/Isolant) ou MIM (Métal/Isolant/Métal): Les modes

SPP peuvent se propager sur des systèmes de type IMI ou MIM. Dans ces struc-

tures, il existe deux interfaces métal/isolant induisant la possibilité d’exciter deux

modes SPP. Les modes liants SPP se propageant à chaque interface peuvent intera-

gir l’un avec l’autre et former des modes couplés quand la distance les séparant est

comparable ou plus petite que la distance de pénétration.

Modes de gap: Lorsque deux particules sont approchées très près l’une de l’autre,

c’est-à-dire à des distances comparables ou inférieures à leur taille caractéristique,

le champ électrique peut être amplifié drastiquement dans la région entre les deux

particules, induisant de que l’on appelle communément un « hot spot ». Des effets

similaires se produisent quand une particule métallique est cette fois placée dans les

environs proches d’une surface métallique.

142

French Summary

Les modes plasmoniques sombres: Il s’agit de modes anti-liants qui, contraire-

ment au mode dipolaire, ont un moment dipolaire net proche de zéro, dont les pertes

d’énergie ont lieu uniquement par dissipation et dont la résonance est beaucoup plus

étroite que les modes clairs.

Effet Fano: Les résonances de Fano présentent un profil asymétrique et ont été

mises en évidence pour la première fois dans des systèmes quantiques. Elles ont pour

origine des interférences constructives et destructives entre une résonance discrète et

étroite et une ligne spectrale large voire continue. Dans les systèmes plasmoniques à

base de nanoparticules métalliques, si plusieurs résonances sont excitées simultané-

ment et si ces résonances se recouvrent spectralement, la signature optique peut

devenir très différente de la résonance plasmonique d’une nanoparticule individu-

elle. Dans le cas où l’une des deux résonances correspond à un mode spectralement

large et l’autre à un mode discret et étroit, les conditions deviennent réunies pour

observer l’apparition de l’effet Fano et obtenir une résonance au profil asymétrique.

Propriétés optiques des systems couples film-nanoparticules:

Les plasmons polaritons de surface (SPP) et les plasmons localisés de surface

(LSP) font l’objet de nombreuses investigations du fait de leur fort potentiel tech-

nologique. Récemment, une attention particulière a été portée à des systèmes sup-

portant les deux types de résonances, SPP et LSP, en déposant des nanopartic-

ules (NPs) métalliques au-dessus de films minces métalliques et en les séparant

par une couche diélectrique. Plusieurs études ont précédemment mis en évidence

le phénomène de couplage et d’hybridation entre les modes localisés et délocalisés.

Cependant, une compréhension en profondeur des propriétés optiques et du potentiel

de ces interfaces est toujours manquante du fait du grand nombre de paramètres

en jeu. Parmi ces paramètres, les plus cruciaux sont certainement la nature et

l’épaisseur de la couche séparatrice comprise entre les deux films métalliques, la

nature et la géométrie des nanoparticules métalliques ainsi que leur distribution,

aléatoire ou ordonnée sous forme de réseau. Dans les toutes dernières années, un

certain nombre d’études ont été menées , plaçant ce sujet dans la bonne direction

afin d’obtenir une compréhension gloable du couplage entre modes localisés et délo-

calisés.

Chapitre 2

Pour comprendre au mieux le travail expérimental réalisé durant ce travail de

thèse, toutes les techniques utilisées sont introduites et détaillées dans ce chapitre,

que ce soit les procédés de nanofabrication ou les techniques de caractérisations

structurales et optiques.

143


Procédés de fabrication:

Les films métalliques et diélectriques ont été fabriqués par deposition physique

en phase vapeur grâce à un évaporateur PLASSYS MEB 4000. Pour la fabrication

des réseaux de nanoparticules, nous avons eu recours à la lithographie électronique

garantissant un contrôle de leur taille et de leur morphologie. Les différentes étapes

de cette technique sont illustrées sur la Fig. 7.1.

Figure 7.1: Description du procédé de lithographie électronique et des étapes as-

sociées.

144

French Summary

Caractérisation structurale:

En optique et nanotechnologie, il est crucial de contrôler la morphologie des

interfaces conçues. Ainsi, la caractérisation structurale doit être impérativement

réalisée. Durant ce travail de thèse, la taille et la forme des nanoparticules ont

été régulièrement contrôlées par microscopie électronique à balayage. Quant à

l’épaisseur et l’homogénéité des films minces, elles ont été contrôlées par toute une

série de mesures d’ellipsométrie. Enfin afin de vérifier la qualité du graphène, nous

avons eu recours à la spectrométrie photoélectronique X (XPS) pour contrôler son

état chimique et à la spectroscopie Raman pour mesurer à la fois sn épaisseur et la

densité de défauts mécaniques.

Caractérisation optique:

Dans ce travail de these, nous avons eu recours à diverses techniques de spec-

troscopies pour étudier les propriétés optiques et plasmoniques de ces interfaces.

Spectroscopie SPR: La spectroscopie SPR est utilisée pour mettre en lumière les

plasmons de surface propagatifs supportés par des films métalliques. Afin d’assurer

les conditions d’accord de phase pour exciter les modes SPP délocalisés, la lumière

incident est propagée à travers un prisme à des angles supérieurs à une valeur critique

garantissant une réflexion totale interne. L’excitation des modes SPP conduit à

l’absorption de la lumière incidente ou à l’atténuation de la lumière réfléchie.

La spectroscopie ultra-violet, visble et infra-rouge (UV-VIS-NIR): Cette tech-

nique de spectroscopie est souvent utilisée pour caractériser les propriétés optiques

de différents matériaux, revêtements, films minces ou solutions. Elle est fondée

sur l’interaction des ondes électromagnétiques avec la matière. Soumis à ce rayon-

nement, les molécules ou les ions contenus dans la matière vont subir des transitions

électroniques qui seront visibles dans le spectre d’absorption.

La spectroscopie Raman: C’est une technique qui repose sur la diffusion Raman,

c’est-à-dire la diffusion inélastique des photons induite par les échanges d’énergie

entre ceux-ci et le milieu qu’ils traversent. Dans le cadre de la spectroscopie Raman,

un faisceau laser, i.e. une lumière monochromatique, est envoyé sur l’échantillon

et l’analyse de la lumière diffusée permet de déduire un ensemble de propriétés

spécifiques à l’échantillon.

145


Chapitre 3

Des recherches autour des systèmes couplés NP/film ont été menés par plusieurs

groupes que ce soit en l’absence ou en présence d’une couche séparatrice qui est

généralement un matériau diélectrique. Cependant, malgré toutes ces études théoriques

et expérimentales, il est toujours difficile de prédire les propriétés optiques de ces

interfaces en fonction de leur design puisque la théorie sous-jacente n’est pas encore

parfaitement établie. Une étude complète, numérique et expérimentale, mettant

en lumière le rôle des différents paramètres géométriques, diamètres et périodicité

des NPs ainsi qu’épaisseur de la couche séparatrice, est toujours manquante. Pour

cette raison, dans les chapitres 3 et 4, nous viserons à apporter une contribution

dans cette direction en menant une étude systématique autour d’interfaces présen-

tant cinq diamètres et trois périodicités possibles de NPs et une épaisseur de couche

séparatrice variant entre 0 et 100nm. Ces interfaces sont constituées de NPs d’or

fabriquées par lithographie électronique et d’une couche séparatrice en SiO2. Nous

avons commencé par comparer les propriétés optiques des réseaux de NPs d’or fab-

riquées directement sur un film d’or à celles de ces mêmes NPs d’or déposées sur des

substrats de verre. Après avoir compris l’influence mutuelle du film d’or et des NPs

d’or sur leurs propriétés métalliques lorsque les dernières sont déposées directement

sur le premier, nous avons ajouté un degré de complexité insérant entre le film et

les NPs une couche séparatrice ultra-minde de SiO2 (6nm). Les propriétés plas-

moniques sont alors profondément modifiées nécessitant une comparaison étroites

entre mesures expérimentales et simulations numériques.

Fabrication des échantillons et caractérisation structurale:

Des réseaux de NPs d’or de 50µm x 50µm de dimension ont été fabriqués par

lithographie électronique. Les différents réseaux étudiés présentent des diamètres

variant de 80nm à 200nm et des périodicités de 300nm, 450nm ou 600nm. Les

NPs ont été fabriquées sur différents substrats : verre, film d’or/verre et couche

séparatrice SiO2/film d’or/verre. La Fig. 7.3 présente des images en microscopie

optique en champ sombre de douze des quinze interfaces étudiées. L’axe des x est

fonction du diamètre tandis que celui des y est fonction de la périodicité des réseaux.

Cette disposition des réseaux a été conservée pour tous les systèmes étudiés durant

ce travail de thèse.

Influence du film métallique sur les propriétés plasmoniques des

NPs métalliques:

Afin de comprendre les changements induits par la presence du film métallique

zn l’absence de couche séparatrice, nous avons comparé deux interfaces avec des

NPs similaires, l’une pour laquelle elles sont fabriquées directement sur un substrat

146

French Summary

Figure 7.2: Schémas des différentes interfaces conçues et étudiées, à savoir des NPs

d ‘Au déposées sur a) des substarts de verre, b) des substrats film d’Au/verre et

c) des substrats SiO2/film d’Au/verre.

de verre et l’autre pour laquelle elles sont fabriquées sur un substrat film d’Au

de 50nm/verre. Comme indiqué sur la Fig. 7.4, les NPs déposées sur un substrat

de verre présentent un mode dipolaire longitudinal qui se décale linéairement aux

grandes longueurs d’onde quand la taille des NPs augmente. Lorsque les NPs d’Au

sont déposées sur un film d’Au, ce mode dipolaire ne peut plus être observé puisqu’il

est décalé à de grandes longueurs d’onde qui ne sont pas accessibles avec notre

spectromètre. Cependant, un autre mode, quasiment indépendant de la taille des

NPs est observé à basse longueur d’onde, vers 520nm. Les simulations numériques

montrent que ce mode correspond à un mode LSP dont les « hot spots » sont

poussées vers le haut de la NP.

Résultats

Afin d’étudier ces structures de type MIM (metal/isolant/metal), nous avons

commencé par déposer une couche de SiO2 de 6nm entre le film d’Au et les NP

d’Au. La Fig. 7.5 présente les spectres d’extinction optique obtenus expérimentale-

ment tandis que la Fig. 7.6 montre les spectres numériques et les cartographies du

champ électrique aux longueurs d’onde de résonance. Ces mesures obtenues sur les

différentes NPS mettent en lumière la richesse de ces interfaces plasmoniques. Ces

structures MIM présentent plusieurs modes plasmoniques:un mode LSP à 520nm dis-

cuté précédemment, un mode SPP délocalisé à l’interface metal/verre de longueur

d’onde 560nm, une résonance Fano entre 650 et 800nm suivant la taille des NPs et

147


Figure 7.3: Image obtenue en microscopie optique en champ sombre de douze

réseaux de NPs d’Au présentant des diamètres de 200nm, 170nm, 140nm, 110nm

(axe x) et trois périodicités, 300nm, 450nm et 600nm (axe y).

enfin un mode LSP longitudinal qui n’est observé que pour les plus petites NPs. Les

résultats expérimentaux concordent parfaitement avec les simulations numériques

menées sur des systèmes similaires.

Les mesures en fonction de l’angle d’incidence permettent d’exciter les modes

SPP qui ne peuvent pas être excités sous incidence normale. Varier cet angle permet

aussi de modifier la position spectrales des modes SPP et par conséquent les condi-

tions d’interaction de la résonance Fano. Cela permet ainsi de coupler et découpler

différents modes plasmoniques et ainsi d’augmenter le degré d’accordabilité de ces

structures MIM.

Chapitre 4

Dans le précédent chapitre, nous avons étudié des systèmes couplés NPs/film et

mis en évidence la richesse de leurs propriétés plasmoniques. Nous avons pu identifier

les différents modes localisés et délocalisés des résonances plasmoniques ainsi que leur

dépendance en fonction de la taille et de la périodicité des NPs ainsi que de l’angle

d’illumination. Dans ce chapitre, nous avons pour objectif d’étudier l’évolution de

ces modes lorsque l’épaisseur de la couche sépératrice diélectrique augmente. Comme

discuté précédemment, divers groupes ont exploré l’influence de couches séparatri-

ces dont l’épaisseur varie entre 20 et 50nm. Pourtant, il manque toujours une étude

complète sur le rôle de l’épaisseur de cette couche séparatrice, c’est ce qui a motivé

les travaux que nous avons menés sur des système dont l’épaisseur de la couche en

SiO2 varie entre 10 et 100nm et que nous présentons ici. Dans ce chapitre, nous mon-

trons l’effet de la couche séparatrice sur les résonances plasmoniques lorsque celle-ci

148

French Summary

Figure 7.4: Spectres d’extinction optique de NPs avec cinq diamètres différents,

80, 110, 140, 170 et 200nm déposées sur a) un substrat de verre et b) un substrat

de verre recouvert d’un film d’Au de 50nm.

augmente, induisant un élargissement, un décalage spectrale voire une disparition de

certaines résonances plasmoniques. Nous mettons aussi en lumière le potentiel de ces

structures MIM comme super-absorbeurs pour les couches séparatrices diélectriques

épaisses.

Evolution des modes plasmoniques lorsque l’épaisseur de lacouche séparatrice augmente

Afin d’étudier l’évolution des modes plasmoniques en fonction de l’épaisseur de

la couche sépératrice diélectrique, nous avons mené une étude comparative entre

mesures expérimentales et simulations numériques. Pour les mesures expérimen-

tales, nous avons effectué des mesures de spectroscopie optique d’extinction. Nous

avons mesuré les spectres d’extinction de NPs sur un substrat et non en solution

ce qui rend la definition de l’extinction optique plus delicate du fait de la réflexion

du substrat.Pour éviter toute ambiguité, l’extinction optique est définie, dans nos

mesures, comme la grandeur –log(T/T0) où T est la transmission mesurée sur les

regions avec NPs et T0 celle mesurée sur les regions sans NPs. C’est la même quan-

tité qui a été analysée dans les simulations numériques que nous avons menées avec

notre collègue Gaëtan Lévêque en recourant à la méthode des tenseurs de Green.

Les résultats expérimentaux et numériques pour les différentes épaisseurs de couche

séparatrice sont présentés sur les mêmes graphes afin d’aider à leur analyse et à leur

interpretation. Ici, les résultats présentés concernent des réseaux de NPs avec une

périodicité de de 300nm et des diamètres de 80, 110, 140 et 200nm. Nous divisons

la discussion antre les petites NPs de 80nm et 110nm et les plus grosses de 140nm

et 200 nm.

149


Figure 7.5: Spectres d’extinction obtenus en transmission et sous incidence nor-

male pour différents diamètres des NPs d’Au (80, 110, 140, 170 et 200nm) avce

des distances centre-à-centre de a) 300nm, b) 450nm et c) 600nm.

Cas des petites NPs Les interfaces MIM sont des systems plasmoniques com-

plexes, présentant à la fois des modes délocalisés et localisés avec des origines très

diverses et un role different de l’épaisseur de la couche diélectrique. La Fig. 7.9 mon-

tre les spectres d’extinction expérimentaux (a) et numériques (b) pour des NPs de

80nm déposées sur un film d’Au recouvert d’une couche en SiO2 d’épaisseur vari-

able. Le mode à 520nm, mis en évidence au chapitre 3, est ici facilement identifiable

pour des épaisseurs faibles de la couche diélectrique mais devient plus difficile à dis-

cerner lorsque l’épaisseur de cette dernière augmente. Concernant le mode SPP à

l’interface métal/verre, il est identifiable à une longueur d’onde de 560nm pour une

faible épaisseur de la couche diélectrique et varie de seulement quelques nanomètres

lorsque la couche diélectrique devient plus épaisse (>60nm). Pour ces épaisseurs

plus élevées de la couche diélectrique, le décalage vers les grandes longueurs d’onde,

certes faible, de la résonance SPP à l’interface métal/verre et le décalage, cette fois-ci

vers les basses longueurs d’onde, du mode LSP longitudinal induit un recouvrement

spectral entre ces deux résonances qui conduit à l’observation d’un seul pic qui peut

expérimentalement résolu. Les résultats expérimentaux montrent un bon accord

avec les simulations numériques.

150

French Summary

Figure 7.6: (a) Spectres d’absorption d’une particule d’Au cylindrique unique pour

différents diamètres déposée au-dessus d’un substrat multicouches . L’excitation

est obtenue par une onde plane incident polarisée TM (transverse magnétique)

sous incidence normale en provenance du côté air. (b) Distribution de l’intensité

du champ électrique pour les modes visibles et indiqués sur le spectre (a). (c)

Amplitude du champ diffusé à l’infini dans la direction de l’onde incidente pour

des particules de 80nm (ligne noire) et 200nm (ligne verte).

Cas des larges NPs Les NPs plus larges offrent un comportement quelque peu

différent. Afin de le mettre en évidence, la Fig. 7.10 présente les spectres d’extinction

optique expérimentaux (a) et numériques (b) de NPs de 200nm. Pour les NPs de

200nm, le mode à 520nm est plus intense et peut être facilement obsrvable jusqu’à

des épaisseurs de la couche diélectrique atteignant 50nm. Le mode SPP à l’interface

métal/verre est aussi facilement identifiable pour les couches en SIO2 allant jusqu’à

80nm. Pour une épaisseur de la couche diélectrique de 100nm, ce mode ne peut plus

être résolu expérimentalement du fait d’un recouvrement spectral avec le mode LSP

longitudinal. Un nouveau mode qui n’avait pas été observé pour les plus petites

NPs apparaît dans les spectres d’extinction et correspond à un mode de gap. Ce

mode apparaît du comportement de la couche séparatrice diélectrique comme une

cavité résonante pour les modes plasmons de surface propagatifs.Ce mode de gap

est particulièrement visible pour des épaissuers de la couche séparatrice diélectrique

de 10nm et ne peut plus être identifié pour des épaisseurs plus élevées du fait que

lorsque la distanec entre NPs et films devients trop grande, la couche diélectrique

ne peut plus agir comme une cavité résonante. Enfin, le mode LSP longitudinal est

visible aux grandes longueurs d’onde. Il apparaît pour ce mode un décalage entre

mesures expérimentales et simulations numériques. En effet, les simulations prédis-

ent un rapide décalage vers les basses longueurs d’onde de cette résonance lorsque

l’épaissuer de la couche séparatrice augmente tandis que les mesures expérimentales

mettent en évidence une résonance large vers 700-750nm, particulièrement pour une

151


Figure 7.7: Evolution des spectres d’extinction en function des angles

d’illumination et de collection pour une polarization a) TE et b)TM.

épaisseur de SiO2 de 100nm. Ces différences peuvent s’expliquer en partie par une

fluctuation de la taille et de la périodité des NPs ainsi que de l’épaisseur de la couche

diélectrique. Une étude numérique plus fine en fonction des conditions d’illumination

montre clairement que l’ouverture numérique de 0,45 de l’illumination explique con-

venablement les résultats expérimentaux et la largeur de cette résonance.

Chapitre 5

Les recherches sur les nouvelles formes de matériaux carbonés ont démarré il

y a trente ans avec la découverte des fullerènes en 1985. Cependant, des travaux

théoriques datant de 1947 avaient déjà suggéré qu’une couche monoatomique de

graphite, aujourd’hui connue sous le nom de graphène, pouvait présenter des pro-

priétés électroniques extraordinaires et notamment un comportement relativiste des

électrons. La fabrication d’un tel matériau était supposée impossible jusqu’à ce qu’en

2004 André Geim, Kostya Novoselov et leurs collègues de l’Université de Manchester

obtiennet par clivage d’un échantillon de graphite hautement pyrolytique à l’aide

d’un sample ruban de scotch. C’est ce qui a marqué le début de « l’âge d’or »

du graphène. Actuellement, plusieurs procédés de synthèse sont accessibles parmi

lesquelles les principales ont l’exfoliation mécanique et le dépôt CVD (Chemical

Vapor Deposition). Le graphène est un plan monoatomique d’atomes de carbone

présentant une structure en nid d’abeille avec une épaisseur de 0,3nm et considéré

jusqu’à présent comme un super « nanomatériau » présentant des propriétés élec-

triques, mécaniques, optiques et thermiques exceptionnelles. Sa stabilité chimique,

sa très grande surface spécifique, sa remarquable transmission optique , plus élevée

que celle de l’ITO (Indium Tin Oxide), son élasticité élevée, sa largeur de bande con-

152

French Summary

Figure 7.8: Spectres d’extinction d’un réseau carré de NPs d’or de 50nm

d’épaisseur,de diamètre 200nm et de périodicité 300nm fabriquées sur un sub-

strat multicouches pour une polarisation incidente TM.Les lignes pleines noires et

rouges permettent de suivre l’évolution des modes SPP à l’interface air (noir) et

verre (rouge).Les lignes vertes indiquent la position spectrale des modes de gap

non-dispersifs. La distribution de l’amplitude du champ électrique à l’intérieur

du plan de polarisation a été tracée pour quelques angles et longueurs d’onde

sélectionnés. Les flèches vertes sont la partie réelle du champ électrique. (b) Com-

paraison entre les spectres d’extinction obtenus pour une polarisation TM et une

polarisation TE en fonction de l’angle d’incidence.

trôlable électriquement et la possibilité de le fonctionnaliser chimiquement ouvrent

un large éventail d’applications notamment dans l’opto-électronique. Les matériaux

plasmoniques plus classiques sont généralement des métaux, plus spécifiquement l’or

et l’argent qui présentent des résonances plasmoniques dans le visible. Du fait des

densités électroniques du graphène, celui-ci présente des résonances plasmoniques

qui se situent dans le domaine infra-rouge. Dans les chapitres précédents, nous

avons montré comment les propriétés plasmoniques de NPs métalliques sont sen-

sibles à la distance les séparant d’un film métallique. C’est précisément ce qui a

motivé cette partie de notre travail, à savoir la réduction de la distance NP /film

métalliques à des distances sub-nanométriques.

Récemment, un certain nombre de groupes a montré l’intérêt que peut revêtir

l’incorporation du graphène dans des dispositifs plasmoniques classiques. Plusieurs

études ont porté sur le potentiel du graphène comme revêtement de films d’Ag ou

153


Figure 7.9: Spectres d’extinction optique a) expérimentaux et b) numériques pour

des réseaux de NPs d’Au de 80nm et de périodicité 300nm déposés sur des films

d’Au recouverts par une couche de SiO2 dont l’épaisseur varie de 10nm à 100nm.

d’Au dans des dispositifs SPR. En effet, intégrer le graphène dans ces dispositifs

présente plusieurs avantages à savoir (1) un ratio surface/voulme élevé permettant

d’élever significativement l’adsorption de biomolécules en comapraison avec des films

métalliques nus, (2) l’adsorption de biomolécules et de molécules organiques du fait

de la structure hexagonale du graphène et des interactions électroniques et (3) la

possibilité de contrôler les résonances plasmons de surface via le nombre de couches

de graphène transférées sur le film métallqiue. Si l’influence du graphène sur les

résonances de systèmes SPR a été caractérisée expérimentalement et numérique-

ment, aucune étude n’a été menée sur son influence lorsqu’il est inséré entre un film

métallique et des NPs métalliques.

Dans la seconde partie de ce travail de thèse, nous avons tiré avantage du

graphène comme couche séparatrice non-diélectrique et à haut indice de réfraction

pour concevoir des systèmes couplés NPs/film avec une couche séparatrice sub-

nanométrique. Pour mieux comprendre ces interfaces plasmoniques hybrides, nous

l’avons comparé à un système similaire sans graphène. Nous avons aussi sondé

l’interaction graphène/NPs métallqiues ainsi que son évolution en fonction du nom-

bre de couches de graphène. Nous avons finalement étudié la possibilité de doper

le graphène en y déposant des NPs métalliques et comment ce dopage du graphène

par les nPs métalliques peut influencer les propriétés plasmonqiues de ce système.

154

French Summary

Figure 7.10: Spectres d’extinction optique a) expérimentaux et b) numériques

pour des réseaux de NPs de diamètre de 200nm et de périodicité 300nm déposés

sur des films d’Au recouverts d’une couche de SiO2 d’épaisseur variant de 10 à

100nm.

Préparation des échantillons et caractérisation

Nous avons fabriqué des structures NPs/film couplés en incorporant du graphène

comme couche séparatrice mais aussi des NP similaires directement sur du graphène

déposé sur un substrat de quartz. Le graphène a été transféré sur ces différentes

interfaces par un transfert à sec comme montré sur la Fig. 7.11. Pour contrôler la

qualité du graphène, nous avons eu recours à la spectroscopie XPS qui a permis de

vérifier que la qualité du graphène n’avait pas été altérée. Nous avons eu effectué des

mesures de spectroscopie Raman afin de mettre en évidence d’éventuels défauts mé-

caniques dans le graphène induits par son transfert, une augmentation de l’intensité

de la bande D donner accès à cette information.

Résonances de plasmons localisés de surface d’interfacesNP/film couplés via une couche séparatrice de graphène

Nous avons étudié ici les propriétés plasmoniques de deux interfaces, l’une avec

des NPs d’or directement déposées sur un film d’or et l’autre pour laquelle des NPs

d’or similaires sont déposées sur un film d’or recouvert de graphène comme montré

sur la Fig. 7.12.

155


Figure 7.11: Illustration du procédé de fabrication d’un des systèmes étudiés à

savoir une interface où les NPs d’Au sont couplées à un film d’or via du graphène

comme couche séparatrice.

La Fig. 7.13 présente les spectres d’extinction obtenus pour les réseaux de NPs

d’or déposés sur les films d’or recouverts ou non de graphène. La présence du

graphène a deux conséquences principales sur les propriétés plasmoniques de ces

structures. D’abord, il y a un léger resserrement du mode à basse longueur d’onde,

ce qui est intéressant pour les applications comme capteurs puisque cela induit une

augmentation du facteur de mérite. Le second effet est un décalage significatif de

13 nm de la résonance aux basses longueurs d’onde par rapport au système sans

graphène. Du fait de l’indice de réfraction élevé, ce décalage de la résonance vers les

basses longueurs d’ondes n’était pas attendu et ne peut pas être expliqué à l’aide

de simulations numériques. L’une des hypothèses que nous avançons est qu’il se

produitun transfert de charge entre le graphène et les NPs métalliques qui modifie

ainsi la fréquence du mode LSP des NPs métalliques via un changement de la densité

des porteurs de charges libres et de la fréquence plasma.

Interaction du graphene avec les structures métalliques

Afin de pouvoir clarifier l’origine du décalage aux basses longueurs d’ondes,

nous avons choisi de réduire la complexité du système en étudiant un système sans

film métallique. Nous avons aussi mené une étude comparative entre des interfaces

156

French Summary

Figure 7.12: Illustration des deux types d’interfaces étudiés: a) réseau de NPs d’or

depose directement sur un film d’or recouvert de graphene et b) réseau similaire

de NPs d’or mais deposes directement sur un film d’or, sans presence de graphene.

quartz/graphène/NPs métalliques à base d’or et à base d’argent. Nous avons pour

cela effectué des mesures de spectroscopie d’extinction et de spectroscopie Raman.

En effet, cette dernière permet de mettre en lumière le dopage du graphene par les

NPs métalliques, les NPs d’Ag induisant un dopage de type n et les NPs d’Au un

dopage de type p. La Fig. 7.14 montre les résultats expérimentaux obtenus sur les

interfaces à base d’Au et la Fig. 7.15 présente ceux obtenus pour les systems à base

d’Ag. On retrouve bien pour l’or un décalage inattendu de la resonance plasmon

vers les basses longueurs d’onde pour les NPs à base d’or quand il y aprésence

de graphene, contrairement à l’argent. Ces résultats mettent en lumière que les

dispositifs hybrides à base de graphene offrent des perspectives intéressantes pour

les applications opto-électroniques, surtout avec la possibilité de doper le graphene

et de contrôle ainsi la resonance des NPs métalliques.

Chapitre 6

Tout au long de ce manuscrit, nous avons étudié et discuté les propriétés op-

tiques de systèmes couplés NPs/Films. Nous avons mis en lumière comment ces

structures MIM possèdent une grande richesse du fait d’un grand nombre de modes

localisés et délocalisés et de leur influence les uns vis-à-vis des autres. Nous avons

aussi étudié la possibilité d’utiliser du graphène comme couche séparatrice plutôt

qu’un matériau diélectrique. Nous avons dans ce chapitre cherché à évaluer le po-

tentiel de ces structures comme nanocapteurs optiques de type LSP ou SERS. Nous

avons pu montrer que ces structures couplées présentent des performances très in-

téressantes en tant que capteurs de type LSP avec des sensibilités élevées et des

largeurs à mi-hauteur des résonances garantissant des facteurs de mérite parmi les

plus élevés de la littérature scientifique dans cette gamme de longueur d’onde. Con-

cernant leur potentiel comme capteurs SERS, nous avons évalué l’exaltation de la

signature Raman du graphène en présence des NPs et nous avons pu montré que

ces structures couplées permettent d’améliorer ce performances en comapraison des

systèmes ne comportant pas de films métalliques.

157


Figure 7.13: Spectres d’extinction d’interfaces avec des NPs d’or déposées directe-

ment sur un film d’or recouvert (lignes pleines) ou non (lignes en pointillés) de

graphène. Les NPs ont un diamètre de 50nm (noir), 80nm (bleu), 110nm (vert) et

140nm (rouge) et une distance centre-à-centre de 300nm. Le signal a été collcté

avec un objectif x10 et une ouverture numérique de 0,15. La référence a été prise

sur une zone de l’échantillon ne contenant pas de NPs.

Capteurs LSP

Afin de sonder le potentiel des structures couplées NPs/film métalliques comme

capteurs LSP, nous avons efectué des mesures expérimentales en fonction de l’indice

de réfraction du milieu environnant. Pour cela, nous avons utilisé des mélanges

eau/glycerol dont l’indice de réfraction varie de 1,33 à 1,47. Nous nous sommes

intéressés en particulier au mode à 520nm qui présente, dans le cas des systèmes

couplés via une couche spérartrice en SiO2 de 6nm Fig. 7.16, un facteur de mérite

de 3,76 ce qui le place parmi les capteurs LSP les plus sensibles reportés dans la

littérature pour cette gamme de longueur d’onde.

Capteurs SERS La spectroscopie Raman exaltée de surface(SERS)

A été mise en évidence de 1977 ouvrant la voie vers la détection de molécules

uniques. Cette exaltation significative est rendue possible par l’apparition de « hot

spots » dans le voisinage de NPs métalliques. Pour des NPs similaires à celles que

nous avons étudiées, l’exaltation du signal Raman du graphène a été évaluée à un

158

French Summary

Figure 7.14: Spectres Raman montrant l’évolution de a) la bande G et b) la

bande 2D du graphene pour des interface graphene/quartz (noir), NPs de

140nm/graphene/quartz (rouge) et NPs de 200nm/graphene/quartz (bleu).c)

Spectres d’extinctions pour des systems NPs/graphene/quartz (rouge) et

NPs/quartz (noir). Un décalage vers les basses longueur d’onde de 13nm est

détecté pour les NPs de 140nm et de 170nm.

facteur de 80 lors de précédentes études. Ici, nous montrons que la présence du film

d’or permet d’atteindre des facteurs d’exaltation de 460 pour la bande G comme

indiqué sur la Fig. 7.17.

159


Figure 7.15: Spectres Raman montrant l’évolution de a) la bande G et b) la

bande 2D du graphene pour des interface graphene/quartz (noir), NPs de

140nm/graphene/quartz (rouge), NPs de 170nm/graphene/quartz (bleu) NPs

de 200nm/graphene/quartz (mauve) .c) Spectres d’extinctions pour des systems

NPs/graphene/quartz (rouge) et NPs/quartz (noir).

160

French Summary

Figure 7.16: a) Spectres d’extinction pour des NPs de 200nm de diamètre et de

300nm de périodicité déposées sur un substrat verre/film d’Au (50nm)/couche

en SiO2 (6nm) en fonction de l’indice de réfraction du mileiu environnant. b)

Décalage du mode LSP à basse longueur d’onde (520nm) en fonction de l’indice

de réfraction du milieu environnant

Figure 7.17: Suivi de l’exaltation de la bande G du graphène pour des sytèmes cou-

plés film d’Au/graphène/NPs d’Au en fonction du diamètre des NPs métalliques.

LA valeur maximale de ce facteur est atteinte pour les NPs de 140nm et vaut 460.

161

Bibliography

[1] Oliver benson. Elements of nanophotonics, May 29 2009.

[2] Plasmonics: Fundamentals and Applications. Springer US, Boston, MA, 2007.

[3] Franz-Philipp Schmidt, Harald Ditlbacher, Ulrich Hohenester, Andreas Ho-

henau, Ferdinand Hofer, and Joachim R Krenn. Dark plasmonic breathing

modes in silver nanodisks. Nano Letters, 12:5780–5783, 2012.

[4] Andrey E. Miroshnichenko, Sergej Flach, and Yuri S. Kivshar. Fano resonances

in nanoscale structures. Reviews of Modern Physics, 82(3):2257–2298, August

2010.

[5] Feng Hao, Yannick Sonnefraud, Pol Van Dorpe, Stefan A. Maier, Naomi J.

Halas, and Peter Nordlander. Symmetry breaking in plasmonic nanocavities:

Subradiant lspr sensing and a tunable fano resonance. Nano Letters, 8:3983–

3988, 2008.

[6] Jonathan A Fan, Chihhui Wu, Kui Bao, Jiming Bao, Rizia Bardhan, Naomi J

Halas, Vinothan N Manoharan, Peter Nordlander, Gennady Shvets, and

Federico Capasso. Self-assembled plasmonic nanoparticle clusters. Science,

328:1135–1138, 2010.

[7] Mario Hentschel, Michael Saliba, Ralf Vogelgesang, Harald Giessen, A Paul

Alivisatos, and Na Liu. Transition from isolated to collective modes in plas-

monic oligomers. Nano Letters, 10:2721–2726, 2010.

[8] Mikael Svedendahl and Mikael Kall. Fano interference between localized plas-

mons and interface reflections. ACS Nano, 6:7533–7539, 2012.

[9] Jiaming Hao, Jing Wang, Xianliang Liu, Willie J. Padilla, Lei Zhou, and Min

Qiu. High performance optical absorber based on a plasmonic metamaterial.

Applied Physics Letters, 96(25):251104, June 2010.

[10] Antoine Moreau, Cristian Ciraci, Jack J. Mock, Ryan T. Hill, Qiang Wang,

Benjamin J. Wiley, Ashutosh Chilkoti, and David R. Smith. Controlled-

reflectance surfaces with film-coupled colloidal nanoantennas. Nature,

492(7427):86–89, December 2012.

[11] Michael G. Nielsen, Anders Pors, Ole Albrektsen, and Sergey I. Bozhevolnyi.

Efficient absorption of visible radiation by gap plasmon resonators. OpticsExpress, 20(12):13311–13319, June 2012.

163

Chapter 7 Bibliography

[12] Mohammad Khaywah. New Ultrasensitive Bimetallic Substrates for SurfaceEnhanced Raman Scattering. PhD thesis, University of technology of troyes,

2014.

[13] Richard P Feynman. There’s plenty of room at the bottom. Engineering andscience, 23, 1960.

[14] K Eric Drexler. Molecular machinery and manufacturing with applications tocomputation. PhD thesis, Citeseer, 1991.

[15] N Tanigushi. On the basic concept of nano-technology proceedings of the

international conference on production engineering tokyo part ii japan society

of precision engineering. 1974.

[16] Lukas Novotny and Bert Hecht. Principles of nano-optics. Cambridge univer-

sity press, 2012.

[17] Harry A Atwater. The promise of plasmonics. Scientific American, 296:56–62,

2007.

[18] Ivan H El-Sayed, Xiaohua Huang, and Mostafa A El-Sayed. Selective laser

photo-thermal therapy of epithelial carcinoma using anti-egfr antibody conju-

gated gold nanoparticles. Cancer letters, 239:129–135, 2006.

[19] Amanda J. Haes and Richard P. Van Duyne. A nanoscale optical biosensor:

a sensitivity and selectivity of an approach based on the localized surface

plasmon resonance spectroscopy of triangular silver nanoparticles. Journal ofthe American Chemical Society, 124(35):10596–10604, September 2002.

[20] Jeffrey N Anker, W Paige Hall, Olga Lyandres, Nilam C Shah, Jing Zhao,

and Richard P Van Duyne. Biosensing with plasmonic nanosensors. Naturematerials, 7:442–453, 2008.

[21] John Brian Pendry. Negative refraction makes a perfect lens. Physical reviewletters, 85:3966, 2000.

[22] Henri J Lezec, Jennifer A Dionne, and Harry A Atwater. Negative refraction

at visible frequencies. Science, 316:430–432, 2007.

[23] Nikolay I Zheludev. A roadmap for metamaterials. Optics and PhotonicsNews, 22:30–35, 2011.

[24] Peter Zijlstra, James WM Chon, and Min Gu. Five-dimensional optical record-

ing mediated by surface plasmons in gold nanorods. Nature, 459:410–413,

2009.

[25] B Lamprecht, JR Krenn, G Schider, H Ditlbacher, M Salerno, N Felidj, A Leit-

ner, FR Aussenegg, and JC Weeber. Surface plasmon propagation in mi-

croscale metal stripes,. Applied Physics Letters, 79:51–53, 2001.

[26] Sergey I Bozhevolnyi, Valentyn S Volkov, Eloise Devaux, Jean-Yves Laluet,

and Thomas W Ebbesen. Channel plasmon subwavelength waveguide com-

ponents including interferometers and ring resonators. Nature, 440:508–511,

2006.

164

Bibliography

[27] AV Akimov, A Mukherjee, CL Yu, DE Chang, AS Zibrov, PR Hemmer,

H Park, and MD Lukin. Generation of single optical plasmons in metallic

nanowires coupled to quantum dots. Nature, 450:402–406, 2007.

[28] Abram L Falk, Frank HL Koppens, L Yu Chun, Kibum Kang, Nathalie

de Leon Snapp, Alexey V Akimov, Moon-Ho Jo, Mikhail D Lukin, and

Hongkun Park. Near-field electrical detection of optical plasmons and single-

plasmon sources. Nature Physics, 5:475–479, 2009.

[29] Sabine Szunerits and Rabah Boukherroub. Sensing using localised surface

plasmon resonance sensors. Chemical communications (Cambridge, England),48(72):8999–9010, September 2012.

[30] J. R. Krenn A. Hohenau. Plasmonic modes of gold nanoparticle arrays on thin

gold films. physica status solidi (RRL) - Rapid Research Letters, 4(10):256 –

258, 2010.

[31] Jack J. Mock, Ryan T. Hill, Aloyse Degiron, Stefan Zauscher, Ashutosh

Chilkoti, and David R. Smith. Distance-Dependent Plasmon Resonant Cou-

pling between a Gold Nanoparticle and Gold Film. Nano Letters, 8(8):2245–

2252, August 2008.

[32] Arash Farhang, Thomas Siegfried, Yasin Ekinci, Hans Sigg, and Olivier J. F.

Martin. Large-scale sub-100Â nm compound plasmonic grating arrays to con-

trol the interaction between localized and propagating plasmons. Journal ofNanophotonics, 8(1):083897–083897, 2014.

[33] Jean Cesario, Romain Quidant, Goncal Badenes, and Stefan Enoch. Electro-

magnetic coupling between a metal nanoparticle grating and a metallic surface.

Optics Letters, 30(24):3404–3406, December 2005.

[34] Yizhuo Chu and Kenneth B. Crozier. Experimental study of the interaction be-

tween localized and propagating surface plasmons. Optics Letters, 34(3):244–

246, February 2009.

[35] Nanfang Yu and Federico Capasso. Flat optics with designer metasurfaces.

Nature materials, 13:139–150, 2014.

[36] Stefan A. Maier and Harry A. Atwater. Plasmonics: Localization and guiding

of electromagnetic energy in metal/dielectric structures. Journal of AppliedPhysics, 98:011101, 2005.

[37] John David Jackson. Classical Electrodynamics Third Edition. Wiley, 1998.

[38] Craig F. Bohren and Donald R. Huffman. Absorption and scattering of lightby small particles. Wiley, 1983.

[39] Gustav Mie. Beitrage zur optik truber medien, speziell kolloidaler metallosun-

gen. Annalen der Physik, 330:337–445, 1908.

[40] Uwe Kreibig and Michael Vollmer. Optical Properties of Metal Clusters.Springer Berlin Heidelberg, 1995.

165


[41] Stephan Link and Mostafa A El-Sayed. Spectral properties and relaxation

dynamics of surface plasmon electronic oscillations in gold and silver nanodots

and nanorods. The Journal of Physical Chemistry B, 103:8410–8426, 1999.

[42] JR Krenn, A Dereux, JC Weeber, E Bourillot, Y Lacroute, JP Goudonnet,

G Schider, W Gotschy, A Leitner, FR Aussenegg, and other. Squeezing the

optical near-field zone by plasmon coupling of metallic nanoparticles. PhysicalReview Letters, 82:2590–2593, 1999.

[43] Matthew Pelton and Garnett W. Bryant. Introduction to Metal-NanoparticlePlasmonics. John Wiley & Sons, 2013.

[44] E Prodan, C Radloff, Naomi J Halas, and P Nordlander. A hybridization

model for the plasmon response of complex nanostructures. Science, 302:419–

422, 2003.

[45] P. Nordlander and E. Prodan. Plasmon Hybridization in Nanoparticles near

Metallic Surfaces. Nano Letters, 4(11):2209–2213, November 2004.

[46] Erin M. Hicks, Shengli Zou, George C. Schatz, Kenneth G. Spears, Richard P.

Van Duyne, Linda Gunnarsson, Tomas Rindzevicius, Bengt Kasemo, and

Mikael Kall. Controlling plasmon line shapes through diffractive coupling in

linear arrays of cylindrical nanoparticles fabricated by electron beam lithog-

raphy. Nano Letters, 5:1065–1070, 2005.

[47] Christy L. Haynes, Adam D. McFarland, LinLin Zhao, Richard P. Van Duyne,

George C. Schatz, Linda Gunnarsson, Juris Prikulis, Bengt Kasemo, and Kall.

Nanoparticle optics:the importance of radiative dipole coupling in two dimen-

sional nanoparticle arrays. The Journal of Physical Chemistry B, 107:7337–

7342, 2003.

[48] B. Lamprecht, G. Schider, R. T. Lechner, H. Ditlbacher, J. R. Krenn, A. Leit-

ner, and F. R. Aussenegg. Metal nanoparticle gratings: Influence of dipo-

lar particle interaction on the plasmon resonance. Physical Review Letters,84:4721–4724, 2000.

[49] J. J. Mock, M. Barbic, D. R. Smith, D. A. Schultz, and S. Schultz. Shape

effects in plasmon resonance of individual colloidal silver nanoparticles. TheJournal of Chemical Physics, 116:6755–6759, 2002.

[50] Xianmao Lu, Matthew Rycenga, Sara E. Skrabalak, Benjamin Wiley, and

Younan Xia. Chemical synthesis of novel plasmonic nanoparticles. AnnualReview of Physical Chemistry, 60:167–192, 2009.

[51] M. Fukui, V. C. Y. So, and R. Normandin. Lifetimes of surface plasmons in

thin silver films. physica status solidi (b), 91:K61–K64, 1979.

[52] Sergey Bozhevolnyi. Plasmonic Nanoguides and Circuits. Pan Stanford Pub-

lishing, 2009.

[53] Dror Sarid. Long-range surface-plasma waves on very thin metal films. Phys-ical Review Letters, 47:1927–1930, 1981.

166

Bibliography

[54] Yuji Kuwamura, Masuo Fukui, and Osamu Tada. Experimental observation

of long range surface plasmon polaritons. Journal of the Physical Society ofJapan, 52:2350–2355, 1983.

[55] Alan E. Craig, Grieg A. Olson, and Dror Sarid. Experimental observation of

the long-range surface-plasmon polariton. Optics Letters, 8:380–382, 1983.

[56] Hiroshi Dohi, Yuji Kuwamura, Masuo Fukui, and Osamu Tada. Long-range

surface plasmon polaritons in metal films bounded by similar-refractive-index-

refractive-index materials. Journal of the Physical Society of Japan, 53:2828–

2832, 1984.

[57] F. Villa, T. Lopez-Rios, and L. E. Regalado. Electromagnetic modes in metal-

insulator-metal structures. Physical Review B, 63:165103, 2001.

[58] J. A. Dionne, L. A. Sweatlock, H. A. Atwater, and A. Polman. Plasmon slot

waveguides: Towards chip-scale propagation with subwavelength-scale local-

ization. Physical Review B, 73:035407, 2006.

[59] Yoichi Kurokawa and Hideki T. Miyazaki. Metal-insulator-metal plasmon

nanocavities: Analysis of optical properties. Physical Review B, 75:035411,

2007.

[60] Lucy H. Smith, Melita C. Taylor, Ian R. Hooper, and William L. Barnes. Field

profiles of coupled surface plasmon-polaritons. Journal of Modern Optics,55:2929–2943, 2008.

[61] Michael J. Preiner, Ken T. Shimizu, Justin S. White, and Nicholas A. Melosh.

Efficient optical coupling into metal-insulator-metal plasmon modes with sub-

wavelength diffraction gratings. Applied Physics Letters, 92:113109, 2008.

[62] Hideki T. Miyazaki and Yoichi Kurokawa. Squeezing visible light waves

into a 3-nm-thick and 55-nm-long plasmon cavity. Physical Review Letters,96:097401, 2006.

[63] Jorge Zuloaga, Emil Prodan, and Peter Nordlander. Quantum description

of the plasmon resonances of a nanoparticle dimer. Nano letters, 9:887–891,

2009.

[64] Jonathan A Scholl, Aitzol García-Etxarri, Ai Leen Koh, and Jennifer A

Dionne. Observation of quantum tunneling between two plasmonic nanopar-

ticles. Nano letters, 13:564–569, 2013.

[65] Kevin J Savage, Matthew M Hawkeye, Rubén Esteban, Andrei G Borisov,

Javier Aizpurua, and Jeremy J Baumberg. Revealing the quantum regime in

tunnelling plasmonics. Nature, 491:574–577, 2012.

[66] Shuming Nie and Steven R. Emory. Probing single molecules and single

nanoparticles by surface enhanced raman scattering. Science, 275:102–1106,

1997.

167


[67] Prashant K Jain, Wenyu Huang, and Mostafa A El-Sayed. On the universal

scaling behavior of the distance decay of plasmon coupling in metal nanopar-

ticle pairs: a plasmon ruler equation. Nano Letters, 7:2080–2088, 2007.

[68] Christopher Tabor, Raghunath Murali, Mahmoud Mahmoud, and Mostafa A

El-Sayed. On the use of plasmonic nanoparticle pairs as a plasmon ruler: The

dependence of the near-field dipole plasmon coupling on nanoparticle size and

shape. The Journal of Physical Chemistry A, 113:1946–1953, 2008.

[69] U. Fano. Effects of configuration interaction on intensities and phase shifts.

Physical Review, 124:1866–1878, 1961.

[70] Boris Luk’yanchuk, Nikolay I. Zheludev, Stefan A. Maier, Naomi J. Halas, Pe-

ter Nordlander, Harald Giessen, and Chong Tow Chong. The Fano resonance

in plasmonic nanostructures and metamaterials. Nature Materials, 9(9):707–

715, September 2010.

[71] Shuang Zhang, Dentcho A. Genov, Yuan Wang, Ming Liu, and Xiang Zhang.

Plasmon-induced transparency in metamaterials. Physical Review Letters,101:047401, 2008.

[72] Feng Hao, Peter Nordlander, Yannick Sonnefraud, Pol Van Dorpe, and Ste-

fan A. Maier. Tunability of subradiant dipolar and fano-type plasmon reso-

nances in metallic ring/disk cavities: Implications for nanoscale optical sens-

ing. ACS Nano, 3:643–652, 2009.

[73] Yannick Sonnefraud, Niels Verellen, Heidar Sobhani, Guy A.E. Vandenbosch,

Victor V. Moshchalkov, Pol Van Dorpe, Peter Nordlander, and Stefan A.

Maier. Experimental realization of subradiant, superradiant, and fano reso-

nances in ring/disk plasmonic nanocavities. ACS Nano, 4:1664–1670, 2010.

[74] Nikolay A Mirin, Kui Bao, and Peter Nordlander. Fano resonances in

plasmonic nanoparticle aggregates. The Journal of Physical Chemistry A,

113:4028–4034, 2009.

[75] Shunping Zhang, Kui Bao, Naomi J. Halas, Hongxing Xu, and Peter Nord-

lander. Substrate-Induced Fano Resonances of a Plasmonic Nanocube: A

Route to Increased-Sensitivity Localized Surface Plasmon Resonance Sensors

Revealed. Nano Letters, 11(4):1657–1663, April 2011.

[76] Kristof Lodewijks, Jef Ryken, Willem Van Roy, Gustaaf Borghs, Liesbet La-

gae, and Pol Van Dorpe. Tuning the Fano Resonance Between Localized and

Propagating Surface Plasmon Resonances for Refractive Index Sensing Appli-

cations. Plasmonics, 8(3):1379–1385, September 2013.

[77] Yuan Yao, Benshun Yi, Jinsheng Xiao, and ZhaoHui Li. Surface Plasmon Res-

onance Biosensors and its Application. In The 1st International Conference onBioinformatics and Biomedical Engineering, 2007. ICBBE 2007, pages 1043–

1046, July 2007.

168

Bibliography

[78] Weihua Zhang, Holger Fischer, Thomas Schmid, Renato Zenobi, and Olivier

J. F. Martin. Mode-Selective Surface-Enhanced Raman Spectroscopy Using

Nanofabricated Plasmonic Dipole Antennas. The Journal of Physical Chem-istry C, 113(33):14672–14675, August 2009.

[79] Kathryn M Mayer and Jason H Hafner. Localized surface plasmon resonance

sensors. Chemical reviews, 111(6):3828–3857, June 2011.

[80] L. R. Hirsch, R. J. Stafford, J. A. Bankson, S. R. Sershen, B. Rivera, R. E.

Price, J. D. Hazle, N. J. Halas, and J. L. West. Nanoshell-mediated near-

infrared thermal therapy of tumors under magnetic resonance guidance. Pro-ceedings of the National Academy of Sciences, 100(23):13549–13554, November

2003.

[81] Lequan Liu, Shuxin Ouyang, and Jinhua Ye. Gold-Nanorod-Photosensitized

Titanium Dioxide with Wide-Range Visible-Light Harvesting Based on Lo-

calized Surface Plasmon Resonance. Angewandte Chemie, 125(26):6821–6825,

June 2013.

[82] Alexandre Aubry, Dang Yuan Lei, Antonio I. Fernandez-DomÃnguez, Yannick

Sonnefraud, Stefan A. Maier, and J. B. Pendry. Plasmonic light-harvesting

devices over the whole visible spectrum. Nano Letters, 10(7):2574–2579, July

2010.

[83] Yoshiaki Nishijima, Lorenzo Rosa, and Saulius Juodkazis. Surface plasmon

resonances in periodic and random patterns of gold nano-disks for broadband

light harvesting. Optics Express, 20(10):11466–11477, May 2012.

[84] Gaeten Léveque and Olivier J. F. Martin. Optical interactions in a plasmonic

particle coupled to a metallic film. Optics Express, 14(21):9971–9981, October

2006.

[85] F. Le, N. Z. Lwin, N. J. Halas, and P. Nordlander. Plasmonic interactions

between a metallic nanoshell and a thin metallic film. Physical Review B,

76(16):165410, October 2007.

[86] N. Papanikolaou. Optical properties of metallic nanoparticle arrays on a thin

metallic film. Physical Review B, 75(23):235426, June 2007.

[87] A. Christ, G. Léveque, O. J. F. Martin, T. Zentgraf, J. Kuhl, C. Bauer,

H. Giessen, and S. G. Tikhodeev. Near-field-induced tunability of surface plas-

mon polaritons in composite metallic nanostructures. Journal of Microscopy,

229(Pt 2):344–353, February 2008.

[88] Amitabh Ghoshal, Ivan Divliansky, and Pieter G. Kik. Experimental observa-

tion of mode-selective anticrossing in surface-plasmon-coupled metal nanopar-

ticle arrays. Applied Physics Letters, 94(17):171108, April 2009.

[89] Jeff DiMaria and Roberto Paiella. Plasmonic dispersion engineering of coupled

metal nanoparticle-film systems. Journal of Applied Physics, 111(10):103102,

May 2012.

169


[90] Gaeten Léveque and Romain Quidant. Channeling light along a chain of

near-field coupled gold nanoparticles near a metallic film. Optics Express,16(26):22029–22038, December 2008.

[91] P-M. Adam T. Maurer and G Leveque. Coupling between plasmonic films and

nanostructures: from basics to applicationsions. Nanophotonics, 2015.

[92] W. R. Holland and D. G. Hall. Surface-plasmon dispersion relation: Shifts

induced by the interaction with localized plasma resonances. Physical ReviewB, 27(12):7765–7768, June 1983.

[93] W. R. Holland and D. G. Hall. Frequency shifts of an electric-dipole resonance

near a conducting surface. Physical Review Letters, 52:1041–1044, 1984.

[94] T. Kume, S. Hayashi, and K. Yamamoto. Light emission from surface plasmon

polaritons mediated by metallic fine particles. Physical Review B, 55:4774–

4782, 1997.

[95] Howard R. Stuart and Dennis G. Hall. Enhanced dipole-dipole interaction

between elementary radiators near a surface. Physical Review Letters, 80:5663–

5666, 1998.

[96] N. Félidj, J. Aubard, G. Lévi, J. R. Krenn, G. Schider, A. Leitner, and

F. R. Aussenegg. Enhanced substrate-induced coupling in two-dimensional

gold nanoparticle arrays. Physical Review B, 66:245407, 2002.

[97] S. Linden, J. Kuhl, and H. Giessen. Controlling the interaction between light

and gold nanoparticles: Selective suppression of extinction. Physical ReviewLetters, 86:4688–4691, 2001.

[98] Dang Yuan Lei, Antonio I. Fernandez-Domanguez, Yannick Sonnefraud, Kan-

natassen Appavoo, Richard F. Haglund, and Pendry. Revealing plasmonic gap

modes in particle-on-film systems using dark-field spectroscopy. ACS Nano,

6:1380–1386, 2012.

[99] Mehdi Keshavarz Hedayati, Mojtaba Javaherirahim, Babak Mozooni, Ramzy

Abdelaziz, Ali Tavassolizadeh, Venkata Sai Kiran Chakravadhanula, Vladimir

Zaporojtchenko, Thomas Strunkus, Franz Faupel, and Mady Elbahri. Design

of a perfect black absorber at visible frequencies using plasmonic metamate-

rials. Advanced Material, 23:5410–5414, 2011.

[100] Min Yan, Jin Dai, and Min Qiu. Lithography-free broadband visible light

absorber based on a mono-layer of gold nanoparticles. Journal of Optics,16:025002, 2014.

[101] Mady Elbahri, Mehdi Keshavarz Hedayati, Kiran Chakravadhanula, Venkata

Sai, Mohammad Jamali, Thomas Strunkus, Vladimir Zaporojtchenko, and

Franz Faupel. An omnidirectional transparent conducting-metal-based plas-

monic nanocomposite. Advanced Materials, 23:1993–1997, 2011.

[102] Zheng-qi Liu, Gui-qiang Liu, Hai-qing Zhou, Xiao-shan Liu, Kuan Huang,

Yuan-hao Chen, and Guo-lan Fu. Near-unity transparency of a continuous

170

Bibliography

metal film via cooperative effects of double plasmonic arrays. Nanotechnology,

24:155203, 2013.

[103] J Britt Lassiter, Felicia McGuire, Jack J Mock, Cristian Ciracì, Ryan T Hill,

Benjamin J Wiley, Ashutosh Chilkoti, and David R Smith. Plasmonic waveg-

uide modes of film-coupled metallic nanocubes. Nano Letters, 13:5866–5872,

2013.

[104] Luna Cui, Gang Song, PeiLin Lang, Chao Wu, Huili Liu, Li Yu, and Jinghua

Xiao. Optical interaction in a plasmonic metallic particle chain coupled to

a metallic film. Optik-International Journal for Light and Electron Optics,124:6936–6938, 2013.

[105] Xiang Wang, Maohua Li, Lingyan Meng, Kaiqiang Lin, Jiamin Feng, Tengx-

iang Huang, Zhilin Yang, and Bin Ren. Probing the location of hot spots by

surface-enhanced raman spectroscopy: toward uniform substrates. ACS Nano,

8:528–536, 2013.

[106] Syed Mubeen, Shunping Zhang, Namhoon Kim, Seungjoon Lee, Stephan

Kramer, Hongxing Xu, and Martin Moskovits. Plasmonic properties of gold

nanoparticles separated from a gold mirror by an ultrathin oxide. Nano Let-ters, 12:2088–2094, 2012.

[107] Sergiy J. Zalyubovskiy, Maria Bogdanova, Alexei Deinega, Yurii Lozovik, An-

drew D. Pris, Kwang Hyup An, W. Paige Hall, and Radislav A. Potyrailo.

Theoretical limit of localized surface plasmon resonance sensitivity to local

refractive index change and its comparison to conventional surface plasmon

resonance sensor. Journal of the Optical Society of America. A, Optics, ImageScience, and Vision, 29(6):994–1002, June 2012.

[108] Carroll F Powell, Joseph H Oxley, John Milton Blocher, and J Klerer. Vapor

deposition. Journal of The Electrochemical Society, 113:266C–269C, 1966.

[109] HJ Levinson. Principles of Lithography. SPIE Press Books, Bellingham, Wash-

ington USA, 2011.

[110] Stanley L Flegler, John William Heckman, Karen L Klomparens, Karen L

Klomparens, and Karen L Klomparens. Scanning and transmission electronmicroscopy: an introduction. WH freeman New York, 1993.

[111] Rasheed MA Azzam and Nicholas Mitchell Bashara. Ellipsometry and polar-ized light. North-Holland. sole distributors for the USA and Canada, Elsevier

Science Publishing Co., Inc., 1987.

[112] Paul Van der Heide. X-ray photoelectron spectroscopy: an introduction toprinciples and practices. John Wiley & Sons, 2011.

[113] Martin P Seah and David Briggs. Practical Surface Analysis: Auger and X-rayPhotoelectron Spectroscopy. john Wiley & Sons, 1990.

[114] Richard BM Schasfoort and Anna J Tudos. Handbook of surface plasmonresonance. Royal Society of Chemistry, 2008.

171


[115] Heinz-Helmut Perkampus, Heide-Charlotte Grinter, and TL Threlfall. UV-VIS Spectroscopy and its Applications. Springer, 1992.

[116] Eric Le Ru and Pablo Etchegoin. Principles of Surface-Enhanced RamanSpectroscopy: and related plasmonic effects. Elsevier, 2008.

[117] C. V. Raman K.S. Krishnan. A new type of secondary radiation. Nature,

121:501–502, 1928.

[118] Interaction of metallic nanoparticles with dielectric substrates: effect of op-

tical constants. Hutter, tanya and elliott, stephen r and mahajan, sumeet.

Nanotechnology, 24:035201, 2013.

[119] Naomi J Halas, Surbhi Lal, Wei-Shun Chang, Stephan Link, and Peter Nord-

lander. Plasmons in strongly coupled metallic nanostructures. Chemical re-views, 111:3913–3961, 2011.

[120] Mathieu L. Juan, Maurizio Righini, and Romain Quidant. Plasmon nano-

optical tweezers. Nature Photonics, 5(6):349–356, 2011.

[121] Peter B Johnson and R-W Christy. Optical constants of the noble metals.

Physics Review B, 6:4370, 1972.

[122] Pierre Viste, Jérome Plain, Rodolphe Jaffiol, Alexandre Vial, Pierre Michel

Adam, and Pascal Royer. Enhancement and quenching regimes in metal-

semiconductor hybrid optical nanosources. ACS Nano, 4:759–764, 2010.

[123] Benjamin Gallinet and Olivier Martin (Electrotechnician). Fano Resonancesin Plasmonic Nanostructures: Fundamentals, Numerical Modeling and Appli-cations. EPFL, 2012.

[124] Andrea Lovera, Benjamin Gallinet, Peter Nordlander, and Olivier J.F. Martin.

Mechanisms of Fano Resonances in Coupled Plasmonic Systems. ACS Nano,

7(5):4527–4536, May 2013.

[125] Patrick C Chaumet, Adel Rahmani, and Garnett W Bryant. Generaliza-

tion of the coupled dipole method to periodic structures. Physical ReviewB, 67:165404, 2003.

[126] Motoichi Otsu. Progress in Nano-Electro-Optics I: Basics and Theory of Near-Field Optics. Springer, 2003.

[127] Gaeten Léveque and Olivier J.F. Martin. Tunable composite nanoparticle for

plasmonics. Optics Letters, 31(18):2750–2752, September 2006.

[128] A Christ, T Zentgraf, SG Tikhodeev, NA Gippius, OJF Martin, J Kuhl,

and H Giessen. Interaction between localized and delocalized surface plas-

mon polariton modes in a metallic photonic crystal. physica status solidi (b),243:2344–2348, 2006.

[129] A Christ, T Zentgraf, SG Tikhodeev, NA Gippius, J Kuhl, and H Giessen.

Controlling the interaction between localized and delocalized surface plasmon

172

Bibliography

modes: Experiment and numerical calculations. Physical Review B, 74:155435,

2006.

[130] NI Landy, S Sajuyigbe, JJ Mock, DR Smith, and WJ Padilla. Perfect meta-

material absorber. Physical review letters, 100:207402, 2008.

[131] Serkan Butun and Koray Aydin. Structurally tunable resonant absorption

bands in ultrathin broadband plasmonic absorbers. Optics express, 22:19457–

19468, 2014.

[132] Harry W Kroto, AW Allaf, and SP Balm. C60: Buckminsterfullerene. Chem-ical reviews, 91:1213–1235, 1991.

[133] KSA Novoselov, Andre K Geim, SVb Morozov, Da Jiang, MIc Katsnelson, IVa

Grigorieva, SVb Dubonos, and AAb Firsov. Two-dimensional gas of massless

dirac fermions in graphene. Nature, 438:197–200, 2005.

[134] Andre K Geim and Konstantin S Novoselov. The rise of graphene. Naturematerials, 6:183–191, 2007.

[135] C emsp14N emsp14R Rao, A emsp14K Sood, K emsp14S Subrahmanyam,

and A Govindaraj. Graphene: The new two-dimensional nanomaterial. Ange-wandte Chemie International Edition, 48:7752–7777, 2009.

[136] Francesco Bonaccorso, Z Sun, T Hasan, and AC Ferrari. Graphene photonics

and optoelectronics. Nature Photonics, 9:622–622, 2010.

[137] A Yu Nikitin, F Guinea, FJ García-Vidal, and L Martín-Moreno. Edge and

waveguide terahertz surface plasmon modes in graphene microribbons. Phys-ical Review B, 84:161407, 2011.

[138] Omer Salihoglu, Sinan Balci, and Coskun Kocabas. Plasmon-polaritons on

graphene-metal surface and their use in biosensors. Applied Physics letter,100:213110, 2012.

[139] Jason C Reed, Hai Zhu, Alexander Y Zhu, Chen Li, and Ertugrul Cubukcu.

Graphene-enabled silver nanoantenna sensors. Nano letters, 12:4090–4094,

2012.

[140] Guowei Xu, Jianwei Liu, Qian Wang, Rongqing Hui, Zhijun Chen, Victor A

Maroni, and Judy Wu. Plasmonic graphene transparent conductors. AdvancedMaterial, 24:OP71–OP76, 2012.

[141] Sabine Szunerits, Nazek Maalouli, Edy Wijaya, Jean-Pierre Vilcot, and Rabah

Boukherroub. Recent advances in the development of graphene-based surface

plasmon resonance (spr) interfaces. Analytical and bioanalytical chemistry,

405:1435–1443, 2013.

[142] Seung Ho Choi, Young L Kim, and Kyung Min Byun. Graphene-on-silver

substrates for sensitive surface plasmon resonance imaging biosensors. Opticsexpress, 19:458–466, 2011.

173


[143] L Wu, HS Chu, WS Koh, and EP Li. Highly sensitive graphene biosensors

based on surface plasmon resonance. Optics express, 18:14395–14400, 2010.

[144] Awnish Gupta, Gugang Chen, P Joshi, S Tadigadapa, and PC Eklund. Raman

scattering from high-frequency phonons in supported n-graphene layer films.

Nano Letters, 6:2667–2673, 2006.

[145] Simone Pisana, Michele Lazzeri, Cinzia Casiraghi, Kostya S Novoselov, An-

dre K Geim, Andrea C Ferrari, and Francesco Mauri. Breakdown of the

adiabatic born–oppenheimer approximation in graphene. Nature materials,6:198–201, 2007.

[146] Anindya Das, S Pisana, B Chakraborty, S Piscanec, SK Saha, UV Waghmare,

KS Novoselov, HR Krishnamurthy, AK Geim, AC Ferrari, et al. Monitor-

ing dopants by raman scattering in an electrochemically top-gated graphene

transistor. Nature nanotechnology, 3:210–215, 2008.

[147] E. Wijaya, N. Maalouli, R. Boukherroub, S. Szunerits, and J.-P. Vilcot.

Graphene-based high-performance surface plasmon resonance biosensors. vol-

ume 8424, pages 84240R–84240R–7, 2012.

[148] RR Nair, P Blake, AN Grigorenko, KS Novoselov, TJ Booth, T Stauber, NMR

Peres, and AK Geim. Fine structure constant defines visual transparency of

graphene. Science, 320:1308–1308, 2008.

[149] Fred Schedin, Elefterios Lidorikis, Antonio Lombardo, Vasyl G Kravets, An-

dre K Geim, Alexander N Grigorenko, Kostya S Novoselov, and Andrea C

Ferrari. Surface-enhanced raman spectroscopy of graphene. Acs Nano, 4:5617–

5626, 2010.

[150] G Giovannetti, PA Khomyakov, G Brocks, VM Karpan, J Van den Brink,

and PJ Kelly. Doping graphene with metal contacts. Physical review letters,101:026803, 2008.

[151] Zheyu Fang, Yumin Wang, Zheng Liu, Andrea Schlather, Pulickel M Ajayan,

Frank HL Koppens, Peter Nordlander, and Naomi J Halas. Plasmon-induced

doping of graphene. Acs Nano, 6:10222–10228, 2012.

[152] Adam M Gilbertson, Yan Francescato, Tyler Roschuk, Viktoryia Shautsova,

Yiguo Chen, Themistoklis PH Sidiropoulos, Minghui Hong, Vincenzo Gi-

annini, Stefan A Maier, Lesley F Cohen, et al. Plasmon-induced optical

anisotropy in hybrid graphene-metal nanoparticle systems. Nano Letters, 2015.

[153] Jonghwan Kim, Hyungmok Son, David J Cho, Baisong Geng, Will Regan,

Sufei Shi, Kwanpyo Kim, Alex Zettl, Yuen-Ron Shen, and Feng Wang. Electri-

cal control of optical plasmon resonance with graphene. Nano letters, 12:5598–

5602, 2012.

[154] Deak A. Kertesz K. Molnar G. Vertesy G. Zambo D. Osvath, Z. and L. P. Biro.

The structure and properties of graphene on gold nanoparticles. Nanoscale,

7:5503–5509, 2015.

174

Bibliography

[155] Hou-Tong Chen, Hong Lu, Abul K Azad, Richard D Averitt, Arthur C Gos-

sard, Stuart A Trugman, John F O’Hara, and Antoinette J Taylor. Electronic

control of extraordinary terahertz transmission through subwavelength metal

hole arrays. Optics Express, 11:7641–7648, 2008.

[156] Hou-Tong Chen, Willie J Padilla, Joshua MO Zide, Arthur C Gossard, An-

toinette J Taylor, and Richard D Averitt. Active terahertz metamaterial de-

vices. Nature, 444:597–600, 2006.

[157] Hou-Tong Chen, Willie J Padilla, Michael J Cich, Abul K Azad, Richard D

Averitt, and Antoinette J Taylor. A metamaterial solid-state terahertz phase

modulator. Nature Photonics, 3:148–151, 2009.

[158] Peter Y Yu, Manuel Cardona, and Lu J Sham. Fundamentals of semiconduc-

tors: physics and materials properties. Physics Today, 50:76, 1997.

[159] Yuanbo Zhang, Yan-Wen Tan, Horst L Stormer, and Philip Kim. Experimental

observation of the quantum hall effect and berry’s phase in graphene. Nature,

438:201–204, 2005.

[160] ZQ Li, Eric A Henriksen, Z Jiang, Zhao Hao, Michael C Martin, P Kim,

HL Stormer, and Dimitri N Basov. Dirac charge dynamics in graphene by

infrared spectroscopy. Nature Physics, 4:532–535, 2008.

[161] Feng Wang, Yuanbo Zhang, Chuanshan Tian, Caglar Girit, Alex Zettl,

Michael Crommie, and Y Ron Shen. Gate-variable optical transitions in

graphene. Science, 320:206–209, 2008.

[162] Jisook Lee, Konstantin S. Novoselov, and Hyeon Suk Shin. Interaction between

Metal and Graphene: Dependence on the Layer Number of Graphene. ACSNano, 5(1):608–612, January 2011.

[163] Eleonora Petryayeva and Ulrich J Krull. localized surface plasmon resonance:

Nanostructures, bioassays and biosensing a review. Analytica chimica acta,

706:8–24, 2011.

[164] Jahan M Dawlaty, Shriram Shivaraman, Jared Strait, Paul George, Mvs Chan-

drashekhar, Farhan Rana, Michael G Spencer, Dmitry Veksler, and Yunqing

Chen. Measurement of the optical absorption spectra of epitaxial graphene

from terahertz to visible. Applied Physics Letters, 93:131905, 2008.

[165] VP Gusynin, SG Sharapov, and JP Carbotte. Sum rules for the optical and

hall conductivity in graphene. Physical Review B, 75:165407, 2007.

[166] Izabela Kaminska, Thomas Maurer, Rana Nicolas, Mikael Renault, Thomas

Lerond, Rafael Salas-Montiel, Ziad Herro, Michel Kazan, J Niedziolka-

Jonsson, Jerome Plain, et al. Near-field and far-field sensitivities of lspr sen-

sors. The Journal of Physical Chemistry C, 119:9470–9476, 2015.

[167] Julia M Bingham, Jeffrey N Anker, Lauren E Kreno, and Richard P

Van Duyne. Gas sensing with high resolution-localized surface plasmon res-

175


onance spectroscopy. Journal of the American Chemical Society, 132:17358–

17359, 2010.

[168] Leif J. Sherry, Shih-Hui Chang, George C. Schatz, Richard P. Van Duyne,

Benjamin J. Wiley, and Younan Xia. Localized Surface Plasmon Resonance

Spectroscopy of Single Silver Nanocubes. Nano Letters, 5(10):2034–2038, Oc-

tober 2005.

[169] T. Maurer, R. Nicolas, G. LÃ©vÃªque, P. Subramanian, J. Proust, J. Beal,

S. Schuermans, J.-P. Vilcot, Z. Herro, M. Kazan, J. Plain, R. Boukherroub,

A. Akjouj, B. Djafari-Rouhani, P.-M. Adam, and S. Szunerits. Enhancing lspr

sensitivity of au gratings through graphene coupling to au film. Plasmonics,pages 1–6, December 2013.

[170] Michelle Duval Malinsky, K. Lance Kelly, George C. Schatz, and Richard P.

Van Duyne. Chain Length Dependence and Sensing Capabilities of the Local-

ized Surface Plasmon Resonance of Silver Nanoparticles Chemically Modified

with Alkanethiol Self-Assembled Monolayers. Journal of the American Chem-ical Society, 123(7):1471–1482, February 2001.

[171] Leif J Sherry, Shih-Hui Chang, George C Schatz, Richard P Van Duyne, Ben-

jamin J Wiley, and Younan Xia. Localized surface plasmon resonance spec-

troscopy of single silver nanocubes. Nano letters, 5:2034–2038, 2005.

[172] Jack J. Mock, David R. Smith, and Sheldon Schultz. Local Refractive Index

Dependence of Plasmon Resonance Spectra from Individual Nanoparticles.

Nano Letters, 3(4):485–491, April 2003.

[173] Sylvia Underwood and Paul Mulvaney. Effect of the Solution Refractive Index

on the Color of Gold Colloids. Langmuir, 10(10):3427–3430, October 1994.

[174] Niels Verellen, Pol Van Dorpe, Chengjun Huang, Kristof Lodewijks, Guy A. E.

Vandenbosch, Liesbet Lagae, and Victor V. Moshchalkov. Plasmon Line Shap-

ing Using Nanocrosses for High Sensitivity Localized Surface Plasmon Reso-

nance Sensing. Nano Letters, 11(2):391–397, February 2011.

[175] M. Grant Albrecht and J. Alan Creighton. Anomalously intense raman spectra

of pyridine at a silver electrode. Journal of the American Chemical Society,

99:5125–5217, 1977.

[176] David L. Jeanmaire and Richard P. Van Duyne. Surface raman spectroelec-

trochemistry: Part i heterocyclic, aromatic, and aliphatic amines adsorbed

on the anodized silver electrode. Chemistry and Interfacial Electrochemistry,

84:1–20, 1977.

[177] Katrin Kneipp, Yang Wang, Harald Kneipp, Lev T. Perelman, Irving Itzkan,

Ramachandra R. Dasari, and Michael S. Feld. Single molecule detection using

surface-enhanced raman scattering (sers). Physical Review Letters, 78:1667–

1670, 1997.

176

Bibliography

[178] M. Kerker, D. S. Wang, and H. Chew. Surface enhanced raman scattering

(sers) by molecules adsorbed at spherical particles: errata. Applied Optics,19:4159–4174, 1980.

[179] Andrea C. Ferrari and Denis M. Basko. Raman spectroscopy as a versatile tool

for studying the properties of graphene. Nature Nanotechnology, 8(4):235–246,

April 2013.

[180] A. N. Grigorenko, M. Polini, and K. S. Novoselov. Graphene plasmonics.

Nature Photonics, 6(11):749–758, November 2012.

[181] P Blake, EW Hill, AH Castro Neto, KS Novoselov, D Jiang, R Yang, TJ Booth,

and AK Geim. Making graphene visible. Applied Physics letters, 91:063124,

2007.

[182] Inhwa Jung, Matthew Pelton, Richard Piner, Dmitriy A Dikin, Sasha

Stankovich, Supinda Watcharotone, Martina Hausner, and Rodney S Ruoff.

Simple approach for high-contrast optical imaging and characterization of

graphene-based sheets. Nano Letters, 7:3569–3575, 2007.

[183] S Roddaro, P Pingue, V Piazza, V Pellegrini, and F Beltram. The optical

visibility of graphene: Interference colors of ultrathin graphite on sio2. NanoLetters, 7:2707–2710, 2007.

[184] Cinzia Casiraghi, Achim Hartschuh, Huihong Qian, S Piscanec, Carsten

Georgi, A Fasoli, KS Novoselov, DM Basko, and AC Ferrari. Raman spec-

troscopy of graphene edges. Nano Letters, 9:1433–1441, 2009.

[185] Yee Kan Koh, Myung-Ho Bae, David G Cahill, and Eric Pop. Reliably count-

ing atomic planes of few-layer graphene (n> 4). ACS nano, 1:269–274, 2010.

[186] Jisook Lee, Sangdeok Shim, Bongsoo Kim, and Hyeon Suk Shin. Surface-

enhanced raman scattering of single- and few-layer graphene by the deposition

of gold nanoparticles. Chemistry -A European Journal, 17:2381–2387, 2011.

[187] Katrin F Domke and Bruno Pettinger. Tip-enhanced raman spectroscopy of

6h-sic with graphene adlayers: selective suppression of e1 modes. Journal ofRaman Spectroscopy, 40:1427–1433, 2009.

[188] Yuika Saito, Prabhat Verma, Kyoko Masui, Yasushi Inouye, and Satoshi

Kawata. Nano-scale analysis of graphene layers by tip-enhanced near-field

raman spectroscopy. Journal of Raman Spectroscopy, 40:1434–1440, 2009.

[189] Wencai Ren, Riichiro Saito, Libo Gao, Fawei Zheng, Zhongshuai Wu, Bilu

Liu, Masaru Furukawa, Jinping Zhao, Zongping Chen, and Hui-Ming Cheng.

dge phonon state of mono-and few-layer graphene nanoribbons observed by

surface and interference co-enhanced raman spectroscopy. Physical Review B,

81:035412, 2010.

[190] Zhenglin Yang, Charity Stratton, Peter J Francis, Mark E Kleinman, Perciliz L

Tan, Daniel Gibbs, Zongzhong Tong, Haoyu Chen, Ryan Constantine, Xian

177

Nomenclature

Yang, et al. Toll-like receptor 3 and geographic atrophy in age-related macular

degeneration. New England Journal of Medicine, 359:1456–1463, 2008.

[191] Cheng-En Cheng, Chi-Yuan Lin, Hao-Yu Chang, Chen-Han Huang, Hsing-

Ying Lin, Chia-Hao Chen, Chia-Chen Hsu, Chen-Shiung Chang, and For-

est Shih Sen Chien. Surface-enhanced raman scattering of graphene with

photo-assisted-synthesized gold nanoparticles. Optics express, 21:6547–6554,

2013.

[192] Xiaoqi Fu, Fengli Bei, Xin Wang, Stephen O’Brien, and John R Lombardi.

Excitation profile of surface-enhanced raman scattering in graphene–metal

nanoparticle based derivatives. Nanoscale, 2:1461–1466, 2010.

[193] Kailin Long, Xiaoguang Luo, Haiyan Nan, Deyang Du, Weiwei Zhao, Zhenhua

Ni, and Teng Qiu. Surface-enhanced raman scattering from graphene covered

gold nanocap arrays. Journal of Applied Physics, 114:183520, 2013.

[194] Jing Niu, Viet Giang Truong, Han Huang, Sudhiranjan Tripathy, Caiyu Qiu,

Andrew TS Wee, Ting Yu, and Hyunsoo Yang. Study of electromagnetic

enhancement for surface enhanced raman spectroscopy of sic graphene. AppliedPhysics Letters, 100:191601, 2012.

[195] Weigao Xu, Nannan Mao, and Jin Zhang. Graphene: A platform for surface-

enhanced raman spectroscopy. Small, 9:1206–1224, 2013.

[196] M Khorasaninejad, SM Raeis-Zadeh, S Jafarlou, MJ Wesolowski, CR Daley,

JB Flannery, J Forrest, S Safavi-Naeini, and SS Saini. Highly enhanced raman

scattering of graphene using plasmonic nano-structure. Scientific Reports, 3,

2013.

[197] Jisook Lee, Konstantin S. Novoselov, and Hyeon Suk Shin. Interaction between

Metal and Graphene: Dependence on the Layer Number of Graphene. ACSNano, 5(1):608–612, January 2011.

178

List of communications:

Journals:

Rana Nicolas, Gaëtan Lévêque, Joseph Marae-Djouda, Guillame Montay, Yazid

Madi, Jérôme Plain, Ziad Herro, Michel Kazan, Pierre-Michel Adam, and Thomas

Maurer, “Plasmonic mode interferences and Fano resonances in Metal-Insulator-

Metal nanostructured interface”, Accepted June 2015, Scientific reports March 16,

2015

Kaminska, T. Maurer, R. Nicolas, M. Renault, T. Lerond, R. Salas-Montiel, Z. Herro,

M. Kazan, J. Niedziolka-Jönsson, J. Plain, P.-M. Adam, R. Boukherroub, S. Szunerits,

“Near-Field and Far-Field Sensitivities of LSPR Sensors”, The Journal of Physical

Chemistry C (JPCC, ACS publications), C 119.17 (2015): 9470-9476.

Thomas Maurer, Rana Nicolas, Gaëtan Lévêque, Palan Subramanian, Julien Proust,

Jérémie Béal, Silvère Schuermans, Jean-Pierre Vilcot, Ziad Herro, Michel Kazan,

Jérôme Plain, Rabah Boukherroub, Abdellatif Akjouj, Bahram Djafari-Rouhani,

Pierre-Michel Adam, Sabine Szunerits. “Enhancing LSPR Sensitivity of Au Gratings through Graphene Coupling to Au Film”. Plasmonics 9, 507–512, 2014.

Proceedings:

Rana Nicolas, Thomas Maurer, Gaetan Lévêque, Palan Subramanian, Julien Proust,

Jérémie Béal, Silvère Schuermans, Jean-Pierre Vilcot, Ziad Herro, Michel Kazan,

Jérôme Plain, Rabah Boukherroub, Abdellatif Akjouj, Bahram Djafari-Rouhani,

Pierre-Michel Adam, Sabine Szunertits, “Enhanced gold film-coupled graphene-based

plasmonic nanosensor”. Proceedings of the SPIE Society 88090-R1, 88090-R7 R.

M.Stockman, 2013.

Thomas Maurer, Aurélien Sarrazin, Alexandre Plaud, Jérémie Béal, Rana Nicolas,

Samuel Lamarre, Julien Proust, Komla Nomenyo, Ziad Herro, Michel Kazan, Gilles

Lerondel, Jérôme Plain, Pierre-Michel Adam, Anna Ritcey. “Strategies for self-organization of Au nanoparticles assisted by copolymer templates”. Proceedings of the

SPIE Society 88092-E1-88092–E8. M. Stockman, 2013.

International Conferences :

R. Nicolas

,G. Lévêque, J. Beal,Z. Herro, M. Kazan, R. Boukherroub, A. Akjouj,P.-M,

Adam, S. Szunerits, T.Maurer “Influence of SiO2 spacer layer on the coupled

plasmonic modes of a gold film- nanoparticles system”, Nanoplasm 2014: “New frontiers in Plasmonics and Nano-optics,” Cetraro, Italy, June 16-20, poster

presentation.

R. Nicolas ,G. Lévêque, J. Beal,Z. Herro, M. Kazan, R. Boukherroub, A. Akjouj,P.-M

Adam, S. Szunerits, T.Maurer, “Interaction between Metallic Nanoparticles and

monolayer graphene, and its effect on LSP resonances”, Nanoplasm 2014: “New frontiers in Plasmonics and Nano-optics,” Cetraro, Italy, June 16-20, oral presentation.

R. Nicolas ,G. Lévêque, J. Beal,Z. Herro, M. Kazan, R. Boukherroub, A. Akjouj,P.-M

Adam, S. Szunerits, T.Maurer, “Influence of the spacer layer dielectric properties on

the coupled plasmonic modes of a gold film-nanoparticles system”, Optical Nanospectroscopy I, Tubingen Germany, March 24-28 2014, poster presentation.

Rana Nicolas, Gaëtan Lévêque, Jérôme Plain, Ziad Herro, Michel Kazan, Pierre-

Michel Adam, and Thomas Maurer, “Coupling and decoupling of localized and

propagating plasmon modes towards exciting Fano resonances in nanostructured

metal-dielectric-metal substrates”, 2nd Optical Nanospectroscopy conference , UcD

Dublin, March 18-20 2015, oral presentation.

Confinement de la lumière dans des métasurfaces plasmoniques nanopar-ticule-film : d'une couche séparatrice d'épaisseur nanométrique à atomique Les plasmons polaritons de surface (SPP) et les plasmons localisés de surface (LSP) font l’objet de nombreuses investigations du fait de leur fort poten-tiel technologique. Récemment, une attention parti-culière a été portée à des systèmes supportant ces deux types de résonances en déposant des nanopar-ticules (NPs) métalliques sur des films minces mé-talliques. Plusieurs études ont mis en évidence le couplage et l’hybridation entre modes localisés et délocalisés. Cependant, une compréhension en pro-fondeur des propriétés optiques et du potentiel de ces interfaces est toujours manquante. Nous avons mené ici une étude de systèmes NPs/film couplés. Nous avons étudié à la fois expérimentalement et théoriquement l’influence d’une couche séparatrice ultra-mince en SiO2 ainsi que l’évolution des diffé-rents modes plasmoniques pour différentes épais-seurs. Nous avons ainsi mis en lumière que de tels systèmes couplés offrent des propriétés optiques exaltées et une large accordabilité spectrale. Nous avons aussi cherché à diminuer l’épaisseur de la couche séparatrice vers le cas ultime monoatomique en utilisant le graphène. Du fait du caractère non-diélectrique de celui-ci, nous avons mis en évidence un comportement optique inattendu de la résonance plasmonique. Nous avons expliqué celui-ci par la mise en évidence du dopage du graphène par les NPs, ce qui est un premier pas en direction de dis-positifs optoélectroniques à base de graphène. En-fin, après avoir amélioré notre compréhension théo-rique de ces systèmes, nous avons évalué leur po-tentiel comme capteurs SERS ou LSP. Mots clés : plasmons - graphène - capteurs optiques - Raman, effet augmenté en surface.

Rana NicolasDoctorat : Matériaux, Mécanique, Optique et Nanotechnologie

Année 2015

Squeezing Light in Nanoparticle-film Plasmonic Metasurface: from Nano-metric to Atomically Thin Spacer Surface plasmon polariton (SPP) and Localized sur-face plasmon (LSP) have attracted numerous re-searchers due to their high technological potential. Recently, strong attention was paid to the potential of SPP and LSP combinations by investigating me-tallic nanoparticles (NPs) on top of metallic thin films. Several studies on such systems have shown the coupling and hybridization between localized and delocalized modes. In this work, we propose a full systematic study on coupled NP/film systems with Au NPs and Au films. We investigate both ex-perimentally and theoretically the influence of an ultra-thin SiO2 dielectric spacer layer, as well as the evolution of the plasmonic modes as the spacer thickness increases. We show that coupled systems exhibit enhanced optical properties and larger tuna-bility compared to uncoupled systems. We also compare these results with those measured for coupled interfaces using graphene as a non-dielectric sub-nanometer spacer. Introducing gra-phene adds complexity to the system. We show that such coupled systems also exhibit enhanced optical properties and larger tunability of their spectral properties compared to uncoupled systems as well as unexpected optical behavior. We explain this behavior by evidencing graphene doping by metallic NPs, which can be a first step towards graphene based optoelectronic devices. After establishing a deep understanding of coupled systems we perform both SERS and RI sensing measurements to validate the high potential of these plasmonic interfaces. Keywords: plasmons (physics) - graphene - Raman effect, surface enhanced – optical detectors.

Ecole Doctorale "Sciences et Technologies"

Thèse réalisée en partenariat entre :

squeezing light in nanoparticle-film plasmonic metasurface

Documents